Nodus 3.1 Manual with Nodus 3.2 Appendix Neuron and network simulation software for Macintosh computers Copyright Erik De Schutter, 1995
Copyright This manual and the Nodus software described in it are copyrighted with all rights reserved. This manual or the Nodus software may not be copied without written consent of Erik De Schutter (E.D.S.), except in the normal use of the software or to make a backup copy. Nodus 3 was compiled with MPW Fortran II, Copyright Absoft Corporation 1991-94. Macintosh is a registered trademark of Apple Computer, Inc. Limited Warranty E.D.S. will replace the media on which this software is distributed at no charge if you report any physical defect within 90 days of purchase and return the item to be replaced. E.D.S. makes no warranty or representation, either express or implied, with respect to this software, its quality, performance, merchantability, or fitness for a particular purpose. As a result this software is sold as is, and you, the purchaser, are assuming the entire risk as to its quality and performance. In no event will E.D.S. be liable for direct, indirect, special, incidental, or consequential damages resulting from any defect in the software or its documentation. The warranty and remedies set forth above are exclusive and in lieu of all others oral or written, express or implied. This manual describes Nodus version 3.1.2. E.D.S. does not guarantee that later versions of Nodus are accurately described by this manual. In the Appendix changes made in Nodus 3.2.1 are described. Information and Service For further information or to report any problem or difficulty with the Nodus 3 software, please write to: Dr. E. De Schutter Born Bunge Foundation University of Antwerp (UIA) Universiteitsplein 1 B2610 Antwerp Belgium Fax 323/820.2541, telex 33646, e-mail erikds@reks.uia.ac.be Reference Papers describing Nodus have been published. Please refer to thes papers when publishing results of modeling with Nodus. E. De Schutter: Computer software for development and simulation of compartmental models of neurons. Computers in Biology and Medicine 19: 71-81 (1989). E. De Schutter: A consumer guide to neuronal modeling software. Trends in Neurosciences 15: 462-464 (1992). E. De Schutter: Nodus, a user friendly neuron simulator for Macintosh computers. in Neural Systems: Analysis and Modeling, F.H. Eeckman editor, Kluwer Academic, Norwell MA. 113-119 (1993). The author would appreciate receiving reprints of any paper referring to work with Nodus.
TABLE OF CONTENTS I. Introduction 5 II. Installing Nodus 3.2 7 III. Modeling with Nodus 9 Introduction 9 The Mathematics of Compartmental Modeling 11 Passive Membrane Models 13 Nodus Implementation of Compartmental Models 15 Excitable Membrane Models 20 Simulation of Synapses and Connections 22 Integration Methods 24 IV. Nodus Reference 27 Nodus Menus 27 Nodus Files 28 Making New Models 36 Making New Simulations 41 Selecting Simulation Parameters 47 V. Nodus Menu Commands 55 Apple Menu 55 File Menu 55 Edit Menu 62 Simulation Menu 64 Network Menu 80 Neuron Menu 83 Conductance Menu 99 VI. Using the Examples 103 Demo Files 103 Realistic Models 112 VII. Appendix 115 Maxima for Memory Storage (Nodus 3.2) 115 Shift and Option Key Menu Modifications 116 Bugs and Problems 117 Import Formats 118 References 120 Nodus 3.2 Update 125 Nodus ftp site 136 VIII. Index 137
I-6 Introduction I. INTRODUCTION There is no Nodus 3.2 manual (yet). This Nodus 3.1 manual has been updated partially to document changes made in Nodus 3.2. This chapter and the next one have been rewritten. However, most of the differences between Nodus 3.2 and Nodus 3.1 are listed in the 'Nodus 3.2 Update' section of the Appendix. A complete rewrite of the manual will be done when additional changes have been made to Nodus. Nodus is a powerful, easy to use application for simulation of neuron and small network models. Compartmental models of neurons with voltage dependent ionic conductances described by Hodgkin-Huxley like equations and with pre- and postsynaptic sites can be created. Nodus runs simulations of these models. Experiments can be performed by injecting different kinds of currents, by simulating voltage clamps or by blocking ionic currents. Simulations can produce a wide variety of color graphics and text outputs. This manual contains all the information necessary to use Nodus 3. It assumes that you have read Macintosh, the owner's guide and are familiar with menus, scrolling, editing text and using the mouse. Read first II. Installing Nodus 3.2, which describes how to install Nodus 3.2 on your hard disk. An introduction to the theoretical aspects of modeling is presented in III. Modeling with Nodus. IV. Nodus Reference describes in depth the Nodus user interface and gives practical instructions about how to use Nodus. V. Nodus Menu Commands has a complete description of each menu command. VI. Using the Examples describes all the example files on the master disk and gives step by step instructions for using them. The Appendix contains some useful lists and references to the modeling literature. Users with no modeling experience should first read chapter III and consult the modeling literature. Refer to chapter VI to try out Nodus with the example files and discover the crucial steps in using Nodus. Every user (including experienced Nodus 2 users) should read chapter IV. The manual uses text attributes to help the user in relating information to the Nodus user interface. All menu commands and dialog window button names and text box titles are printed bold. All Nodus file types are printed italic when they are mentioned for the first time in a paragraph. Warnings are underlined. Titles of manual chapters and file names are quoted. If you want more information on Nodus 3 or if you have any comments or suggestions please contact the author. User feedback helps me to adapt future versions of Nodus better to the needs of neurobiologists. Please do not distribute copies of this program; it contains copyrighted software. If you have obtained a free copy you can still send your name and address to the author, you will be added to the Nodus mailing list and receive information on updates, etc.
Introduction I-7
II-8 Setting up Nodus II. INSTALLING NODUS 3.2 The required hardware is a Macintosh computer with a 68020 or 68030 microprocessor and a numeric coprocessor (FPU) or a 68040 microprocessor having at least 4 megabytes (Mb) of memory and running System 6 or 7. You will get optimal performance using System 7. Nodus 3.2 requires at least 2.0 Mb free RAM to run, it has been preset to use 2.2 Mb. The size of the application heap memory mainly determines how may windows can be shown on the screen simultaneously. Simulation plot windows with color graphics can take up a lot of memory (up to 0.1 Mb). If available memory gets low, Nodus will warn the user and suggest to close some windows. Nodus 3.2 comes with a Nodus_Preferences file. Nodus 3 cannot run without this file. Nodus cannot make a default preferences file so be sure to have a backup! Note that the preferences file has been personalized for each registered user, do not mix files from different registered users. The Nodus Preferences file should be placed together with the Nodus 3.2 application file (i.e. both on the desktop or both in the same folder) or it can be placed in Preferences folder (System 7) in the System Folder. Nodus 3.2 is available in two versions: the standard version (Nodus 3.2) which runs on any Mac with 68020/30/40 CPU and a FPU and the Quadra version (Nodus 3.2Q) which runs only on Macs with a 68040 CPU (it is considerably faster than 3.2). Nodus does not run on Macs without a floating point unit (FPU) or with a disabled FPU (like the 68LC040). However, if you want to test Nodus, the Nodus ftp-site has a software patch SoftFPU that can replace the FPU (this is very slow though, so you want to upgrade your Mac if you are going to use Nodus extensively). You are allowed to use either one or both versions of Nodus. Both versions are completely file compatible with each other and use the same Nodus_Preferences file. Both versions are also 32-bit clean and compatible with the 68040 cache on the Quadras (using the cache results in faster performance).
Setting up Nodus II-9
III-10 Modeling III. MODELING WITH NODUS Introduction 9 The Mathematics of Compartmental Modeling 1 1 Linear cable theory...11 Compartmental models...11 Equivalent circuit...11 Membrane surface...13 Passive Membrane Models 1 3 Morphology...13 Cable parameters...14 Nodus Implementation of Compartmental Models 1 5 Reduced models...15 Weight factors...16 Node and branch connections... 16 Tree format model...17 An example of a model reduction...18 Excitable Membrane Models 2 0 Hodgkin Huxley equation...20 Equivalent circuit...21 Simulation of Synapses and Connections 2 2 Presynaptic transmitter release...22 Postsynaptic conductance...23 Connections between neurons...24 Integration Methods 2 4 Hybrid Euler method...25 Fehlberg method... 25 Experiments...26 Introduction This chapter describes how Nodus models neurons and small networks. It introduces the underlying mathematics and has some practical suggestions about making compartmental models. Modeling methods particular to Nodus are defined. Finally it specifies the different integration methods that are available. This chapter does not pretends to be a complete cookbook about modeling. Different approaches to modeling are possible. The big categories are single cell versus (small) network simulations and passive membrane versus excitable membrane models. All these models describe the electrical status of neurons. There is now a lot of interest in adding (limited) simulations of chemical events to these models, for example the movement and concentration of calcium ions. Which type of model is selected depends on the available experimental data and the particular interests of the modeler. An important issue in creating a model is the specification of the model parameters. An ideal model would have all its parameters based on hard experimental data. Usually only subsets of data are available, so that several parameters need to be estimated. Different model categories contrast in which model parameters are emphasized.
Modeling III-11 In passive membrane models detailed morphology is usually the key issue, while the abstraction is made that for the studied phenomena voltage dependent ionic currents are not critical. Excitable membrane models emphasize the ionic currents, with often a very simple morphology of the neuron (one compartmental models). In most network models connections between the neurons are more important than their morphology or their membrane properties. Nodus is suitable for simulations of all these model categories, but in always a realistic, biological approach is emphasized. All parameters have to be specified by the user, Nodus does not implement random variation of parameters. In the near future new modeling options will be added, including ion concentrations. The consensus in the modeling community about what are the characteristics of a good model evolves. The aspirant modeler should consult the recent literature, which has become quite extensive. See the Appendix for a list of good introductory papers and books. Figure III/1: compartmental models (right) can preserve the detailed morphology of the original neuron (left).
III-12 Modeling The Mathematics of Compartmental Modeling Nodus uses compartmental models to simulate the electrophysiology of neurons. The compartmental approach is easy to use and makes accurate modeling of both anatomy and electrophysiology possible. Linear cable theory The linear cable theory makes exact mathematical formulations of neurophysiological events possible, but its practical use for realistic modeling is limited. Linear cable theory describes the electrophysiological events in a neuron by a single differential equation and for every electrical event specific solutions of this equation have to be obtained. The complexity of these solutions is very sensitive to the complexity of the model so that only simple models (e.g. ball and stick neuron model, 3/2 power branching, etc.) can be handled in practice. Linear cable models are often inadequate to simulate the behavior of real neurons with local interactions among dendrites, inhomogeneous distribution of synapses and ion channels, etc. It is also difficult to incorporate non linear behavior (like voltage dependent ionic currents) in linear cable models. Compartmental models Compartmental modeling is derived from linear cable theory. The basic assumptions are that the neuron is a system of connected membrane cylinders in which the intracellular current flow is essentially parallel to the cylinder axis, and that the resistance of the extracellular medium is negligible. Instead of describing the whole neuron by one large and complex equation, the neuron is divided into many small parts called compartments and the electrophysiology of each separate compartment is described by a single equation. This equation is simple because the compartments are kept small enough to be considered isopotential (i.e. the membrane voltage is constant over the whole compartment). For passive compartments (i.e. a compartment without voltage dependent ionic currents) the equation is always the same, for excitable compartments terms describing the ionic conductances have to be included. The first step in compartmental modeling is to divide the neuron in compartments, for mathematical reasons only cylinders and spheres are used. It is evident that accurate modeling of morphological details is possible, though it will not always be necessary to make the model as complex as in the example in Fig. III/1. All the compartments in the example, except the soma, are cylindrical with a length L and a diameter D: D L Figure III/2: a cylindrical compartment with length L and diameter D and bilateral connections. Equivalent circuit The compartment can be reduced to an electronic circuit which controls the membrane potential (Fig. III/3). The equivalent electronic circuit consists of a part describing the current flowing through the cell membrane, with a membrane capacitance (CM) and a potential source (E) coupled to a membrane resistance (RM), and a part describing the current flow to other compartments over a cytoplasmic resistance (RI). The values of CM, RM and RI depend on the size of the compartment and on neuron specific equivalent cable parameters (C m, R m and R i ). CM = C m π DL Eq. 1
Modeling III-13 RM = RI = 4R i L R m π DL Eq. 2 πd 2 Eq. 3 RI E CM RM Figure III/3: the equivalent circuit for an isopotential compartment. The potential source E in a passive compartment is the difference between the membrane potential and the resting membrane potential, in an excitable compartment it is made up of several ionic conductances coupled in parallel (see Fig. III/7 and Eq. 28). The change in membrane potential ( E) in a compartment #n with a synaptic current I s and an injected current I e can be described by a differential equation: E n E k E n = E n E r + k RI nk + I s + I e t RM n CM n CM n CM n CM n Eq. 4 The first term in equation 4 describes the current flowing through the cell membrane when the membrane potential is not equal to the resting membrane potential (E r ), the second term is the current flow to other compartments (#k) and the last two terms are the synaptic and injected currents. Each compartment in the model can be described by this simple equation: only the values for E, RM, CM, RI and k change. Mathematically compartmental modeling is simple because the same equation is always repeated. The complexity of the neuron model is contained in the size specific parameters RM, CM and RI, in the number of compartments and the connections between them. Nodus contains a loop which calculates equation 4 for each compartment (the parameters are stored in arrays), this loop is entered at least once during each integration step. From the size specific parameters RM, CM and RI two important values can be derived: the time constant τ m and the space constant λ for the compartment. The time constant τ m for the simple capacitive circuit describing a compartment is given by: τ m = RM i CM i or τ m = R m C m Eq. 5 The space constant λ for a cylinder is the length of a cable with the same diameter D that has RM equal to RI:
III-14 Modeling λ = R m D n 4R i Eq. 6 The electrotonic length l n of a compartment is its length relative to the space constant λ: l n = L n λ Eq. 7 The electrotonic length l n of a compartment is an important measure in determining the electrical accuracy of the model, if it is too large the compartment cannot be considered isopotential. Another important electrophysiological parameter for neurons is the input resistance R N (Eq. 8). It is measured by injecting a steady current I in the cell and it depends both on the local morphology (different R N in the soma compared to the dendrites) and the cable parameters. R N = E I Eq. 8 Membrane surface Membrane surfaces for the spheres and (hollow) cylinders (Eq. 1,2). are computed by standard geometric equations. For spherical compartments the holes in the membrane surface caused by connections to dendritic or axonal cylinders are subtracted from the computed surface. For cylindrical end compartments (at the tip of a dendrite or axon) the surface of the closed end is added to the surface of the cylinder. Inaccuracies in measurements of the size of compartments (due to shrinkage or membrane folding, see next section) usually underestimate the membrane surface. In Nodus corrections can be made for underestimation of membrane surface at 2 levels. For correction of known measurement errors one can Scale Sizes of all compartments (Neuron menu); lengths and diameters can be scaled separately. The changes in compartment lengths and diameters will affect the values for RM, CM and RI. An alternative solution, appropriate to compensate for membrane folding, is the use of a global scaling factor SF which changes for all compartments CM and RM, while leaving RI and τ m unchanged: RM ' = RM CM ' = SFCM SF RI ' = RI Eq. 9 Passive Membrane Models Passive membrane models do not contain voltage dependent ionic currents. The neuronal membrane is considered linear (i.e. the input resistance is constant) over the voltage range that is simulated. They are the most popular type of model, partially because most of the needed parameters are easy to obtain. Morphology Theoretically the best passive membrane model would be one that is a detailed morphological equivalent of the simulated neuron, with all the cable parameters measured in the same cell. In a lot of preparations it is difficult or impossible to get all these data out of one cell, therefore an acceptable approach is to combine accurate morphology of one neuron (obtained from camera lucida measurements on a neuron labeled by intracellular injection of HRP, Lucifer Yellow, etc.) with average physiological data from several neurons.
Modeling III-15 Often some of the cable parameters have to be found by trial and error: several values are used in the model till experimental measurements can be replicated during simulation. Sometimes one is confronted with the problem of multiple possible model parameter values; where one cannot determine from the available experimental data which of these parameter values is the real one. This problem has been reported for a limited number of neuron models only. One can argue that it does not make much sense to use a detailed morphological model if average physiological data are used. Usually there is a great variance in morphological details of branching (from second or third order on) between functionally equivalent cells, both in vertebrates and in invertebrates and the branching pattern in some neurons may change over time. The accuracy of morphology may be illusory. Because the measurements are done on fixed and dehydrated material, the neuron has shrinked and the amount of shrinkage can be different for diverse parts of the cell. The shrinkage can be estimated by measuring some easy to recognize structures before (for example on photographs of a Lucifer Yellow labeled cell) and after preparation but the results will be approximate. Computation time is related to the size and complexity of a model and simulations of large, detailed models may take a lot of time on microcomputers. It makes sense to reduce the size of the model to increase calculation speed, but there is no golden rule on where the best balance between accuracy and computation time is situated. A good approach is to design first a detailed model, then reduce it to about 30 to 150 actual calculated compartments (see the next section on how to do this) and compare the reduced model and the detailed model for important physiological parameters as input resistance, time constants and attenuation of membrane potentials. Increase the complexity of the reduced model if necessary, or reduce it further if possible. Compartments should be neither too large nor too small. Compartments are supposed to be isopotential, this puts an upper limit on their size. It would seem that very small compartments give a greater accuracy, but this is not true in computer simulations because extremely small (or large) numbers give larger round off errors and may even cause numeric overflow (i.e. the number is too large to be represented in the numeric format used by the computer). Try to keep connected compartments to a similar size, large differences in size between connected compartments may also give larger calculation errors. A good measure for the electrical size of a compartment is its electrotonic length (see previous section) which should be between 0.200λ and 0.015λ. Cable parameters The delicate spot in most passive membrane models are the equivalent cable parameters; in a lot of cases their values cannot be measured experimentally. Most authors take for specific membrane capacitance the universal value of 1 µf/cm2, though one can find measured values of 0.3 to 5 µf/cm2. The discrepancies in these measurements are probably due to underestimation of the membrane surface. Specific membrane resistance varies between different species and between different neurons in an animal. Membrane resistance has to be defined for passive compartments, in excitable compartments it is the inverse of the sum of all voltage dependent ionic conductances. One should try to find an exact value for R m. Two approaches can be used: either R m is calculated from time constant data (Eq. 5) or R m is found by trial and error from input resistance measurements in the model. No matter which method is used, one should always use the other one as a final check on the accuracy of the parameter value. To calculate the time constant accurate measurements of the passive response to a hyperpolarizing current step or the relaxation phase after a depolarizing pulse should be made. The choice of depolarization versus hyperpolarization depends on the nonlinear behavior of the cell and R N : stay within a linear range of membrane potential.
III-16 Modeling The potential response is usually determined by several time constants, these can be found by the exponential peeling method. Some time constants may depend on activation or inactivation time constants of ionic currents instead of the shape and cable parameters of the cell. R m can be determined from the slowest time constant τ m of the neuron (Eq. 5). Another method is to find R m with simulations. The input resistance at the soma should be measured in the original preparation in the linear range of membrane potential. If the electrode leak was small this value is mainly determined by the size and shape of the cell (which are supposed to be replicated accurately in the model) and the membrane resistance. R N for the neuron model is measured in simulations with different values for R m till a good approximation is found. If R N can also be measured at others sites (axon, dendrites) one should do so. This method is superior to using exponential peeling data if R m is not constant in the cell. Some authors found that to get an accurate model, R m had to be much larger in the dendrites than in axon and soma (reflecting differences in local ionic conductances). Nodus supports variable membrane resistance in a neuron model. Cytoplasmic resistance can be measured if the axon or a similar structure can be penetrated at two different sites to measure the attenuation of an injected current pulse between these two positions. Often the neuron is too small to be sticked at two sites and even if the experiment is possible the actual measurements may be made inaccurate by bridge imbalances, etc. Most authors therefore use values obtained in other neurons of the same animal or of the same phylogenetic class. Simulations should be done for a range of cytoplasmic resistances if no measured values are available. Nodus allows free mixing of passive and excitable compartments in a neuron model. Any compartment can also have postsynaptic currents and/or presynaptic transmitter release. Nodus Implementation of Compartmental Models Nodus uses an experiment look-a-like approach to modeling neurons and emphasizes the importance of morphology. This is implemented by placing the nodes of the equivalent circuits at the center of the compartments (Fig. III/4) and by using alternative methods to lump branches together. 1 2 Figure III/4: equivalent circuit for a node connection between 2 cylindrical compartments. Reduced models The computation time for a simulation depends on several factors: the integration method, precision of the method, size of the model and total number of conductances. One can simplify the model of a neuron to keep the number of compartments (and the computation time) reasonable. The classic way to simplify dendritic trees is to lump branches together into an equivalent cylinder, i.e. to use one large dendrite instead of several small ones. An extreme example of this method is the ball and stick model used in linear cable models: one spherical soma compartment and a large cylindrical compartment equivalent to the axon, the dendrites, etc.
Modeling III-17 Alternative methods are used in Nodus to decrease the number of compartments with a constant apparent morphological detail: weight factors and branch connections. These techniques are conceptually simpler and make switching between reduced and complex models easy. Nodus can make reduced models automatically. While these methods can make models more manageable, one should realize that there is always a loss in accuracy. Inexperienced users should not use weight factors or branch connections; one make nice models in Nodus without ever using them. Nodus defaults always to (standard) node connections and a neutral weight factor (of one). Weight factors Weight factors are a way to lump branches together without changing their size. Instead of lumping n branches together into one huge branch, an average branch is made and connected n times to the parent compartment (Eq. 10-13; Fig. III/6); n is the weight factor. All compartments in the averaged branch are calculated only once by Nodus; there can be no difference in membrane voltage between the n branches. Mathematically this is similar to lumping branches together, the advantage is that a normal", morphological size of the branch is used. Node and branch connections Branch connections are used to increase the electrical accuracy for some connections in the model, without increasing the number of compartments. In a classic compartmental model all compartments are connected by node connections: the end of one cylinder is connected to the end of the next cylinder, etc. (Fig III/4) If several compartments connect together at one node (for example where a branch splits), each compartment is connected to all other compartments. RI node, 1 2 = 2 R i n L 1 πd 1 2 + L 2 2 πd 2 Eq. 10 RI node, 2 1 = nri node, 1 2 Eq. 11 1 2 Figure III/5: equivalent circuit for a branch connection between 2 compartments.
III-18 Modeling If more than 2 compartments are connected by nodes to the same parent compartment, there is electrotonically no difference between a binary branch (the parent splits into 2 similar sized branches, typical for vertebrate neurons) or a continuation of the parent branch (Nodus assumes this for the first connection) and a smaller side-branch (invertebrate neurons). If the number of compartments in a parent branch is reduced, the accuracy of the model is decreased because several side-branches, which do not originate from the same point on the parent branch, will have to be connected to the same node. Branch connections connect to the center of the cylindrical compartment of the parent branch instead of to the end (Fig. III/5). If several compartments make branch connections to the same compartment they are not connected to each other. Current flow has to pass through the parent branch first, which is more accurate if the compartments are equivalent to branches which do not connect to the same point on the parent branch. RI branch, 1 2 = 2 R i n L 2 2 πd 1 Eq. 12 RI branch, 2 1 = nri branch, 1 2 Eq. 13 Branch connections allow more accurate modeling of complex branching from a lumped parent branch. For each parent compartment connections can be made at two sites (as a node and as a branch), with a distance of halve the length of the compartment between them. To use branch connections the tree format model option has to be on. Branch connections are not necessary to use compartmental models; node connections are sufficient. Use branch connections only if needed and if you understand the concept. They are particularly useful in modeling invertebrate neurons were a lot of thin dendrites originate close to each other from a thick neurite. They can also be used to model dendritic spines. They are less useful in neurons were dendrites show binary branching (as in vertebrate neurons). Tree format model The tree format option for neuron models helps in automatically creating and checking connections between compartments. A tree format neuron model has to be to defined in a centrifugal way. First the soma is defined, then branches originating from the soma, etc. In other words: if each division increases the order of the more distal sections of the branch, then the low order compartments should be defined before the high order compartments. The user is free to define either all compartments of a main branch first and then the compartments of its side-branches, or to define the compartments of a complete side-branch before the more distal parts of the main branch are defined. If the tree format model is selected Nodus enforces some simple rules for the neuron model: - compartments can be connected by only one connection. - a spherical compartment cannot be connected to another spherical compartment. - all connections and their weight factors are defined at the parent compartment (the compartment proximal to the soma, having the lowest order of branching). Connections can be changed or deleted only in the parent compartment definition window. - connecting compartment #n to #k automatically connects #k to #n. - if compartment #k and #l are connected to the cylindrical compartment #n by node connections, then #k and #l are also connected to each other by a node connection ( cross connection ). Node connections to spherical compartments are not interconnected. - compartments connected as a branch can have only one parent compartment. - reverse weight factors (i.e. child to parent) are always equal to one.
Modeling III-19 An example of a model reduction The goal of simplifying a model is to reduce the number of compartments to a minimum with respect for significant morphological details like the number of side-branches, the distance between their point of origin and their cross section and membrane surface. The use of weight factors and of branching connections to obtain this goal is demonstrated in Fig. III/6. Fig. III/6 shows a part of a neuron model: a branch with 5 side-branches, connected to a larger branch through compartment C. This model is a simple example: usually the side-branches themselves will also consist of several compartments. In the original model all connections are node connections and 13 compartments are necessary to describe the complete branch. The first step in making a reduced model is to lump the compartments of the parent branch together. The choice of which compartments are to be lumped depends on two factors: the diameter of the compartments should be comparable and the distance between the branches connected to these compartments should be either small or have no influence on the simulated electrophysiological events. In this example the choice is simple because some compartments have already a common diameter; 3 compartments can be fused to a long compartment D in and 4 small compartments to a compartment E (Fig. III/6B). If the diameters of the compartments selected for lumping are not identical, an average diameter has to be calculated (Eq. 14-17). A A F G C D E B B A F G n B b b b b b C D E n n C A F G 2 3 C D E B Figure III/6: Successive steps in reducing a branch of a compartmental model (A). First the main branch is lumped together and all node connections (n) are converted into branch connections (b) (B). Then the small branches are lumped together and connected with weight factors (C).
III-20 Modeling The connections between compartments C and D and between D and E remain node connections (n). The connections between the side-branches and the parent branch (F to D and G to E) become branch connections (b). The position on the parent branch is more accurate than it would be with a node connection at the end of the respective parent compartments, but the distance in origin between the 2 side-branches F and between the 3 side-branches G has disappeared. The effect of this reduction on the electrical accuracy of this model is limited because the branches F and G are small and far removed from the soma of the model (interposition of several compartments: D, C, A, ). The next step is to lump the side-branches together into one, average branch and to connect this branch n times to the parent branch; the 2 branches F and the 3 branches G are lumped together (Fig. III/6C). The size of the lumped compartment is not just the average of the lengths (L) and diameters (D) of the original compartments. The important electrical sizes are the membrane surface of the compartment (for CM and RM) and its length divided by cross section (for RI). Calculate first the surface (S) and length divided by cross section (C) of all the (cylindrical) compartments: S = πld Eq. 14 4L C = πd 2 Eq. 15 Calculate the averages of the surfaces (S a ) and cross sections (C a ) and use these to calculate length (L a ) and diameter (D a ) of the lumped compartment: L a = S a πd a Eq. 16 4S a D a = 3 π 2 C a Eq. 17 An important characteristic of branch connections is their asymmetry. From parent compartment D it looks like there are 2 child compartments F, but from F there is only one D. This gives a weight factor of 2 for D->F and a weight factor of 1 for F->D. For all connections only the weight factor from parent to child has to be supplied, the reverse weight factor is calculated and controlled by Nodus. The use of branch connections and weight factors has reduced the number of compartments in this part of the example from 13 to 5! Of course there is a decrease in morphological accuracy in the reduced model. In most simulations this will not matter. As long as the branch beginning at compartment C is completely passive and the membrane voltage is monitored in compartments connected to A or B, the effect of the reduction on for example R N will be small. The reduction can be acceptable for current injections (as for example in synapses) in all the side-branches F and/or G together, if membrane potential is monitored in compartments connected to A or B: the effect of the reduction will be greater, but in most cases not substantial. The reduction is not acceptable if one monitors membrane voltage in one of the reduced compartments or if one wants to study local interactions between side-branches F and G. The example shown here is reproduced in the neuron models Test-cell 5a, Test-cell 5b and Test-cell 5c on the Nodus Master disk. Using the examples shows numerical results for current injections in these models. It is important to check the accuracy of a reduced model. The best way to do this is to compare it with the original, complex model, but if this is not available one should compare with a model that has at least twice the number of compartments.
Modeling III-21 Compare the reduced and complex model by repeating simulations of the relevant experiments. Significant values are in most cases values that can be compared with biological measurements like the R N and τ m measured in the soma (in some models other structures like a large axon may be more important) or size and timing of action potentials and synaptic events. The acceptable relative error will vary, but in most cases errors of up to 5% will be no problem (one compares with biological events which show even greater variability). Another way to keep reduced models to an acceptable level of accuracy, is to start with the complete, complex model and then reduce it progressively till the error becomes too large. Excitable Membrane Models Compartments can have excitable membrane in Nodus. Any voltage dependent ionic conductance that can be described with Hodgkin-Huxley like equations can be included. The ionic conductances are defined apart from the neuron model definition. Hodgkin Huxley equations Hodgkin and Huxley based their equations on a model of the conducting channel that consists of activation and inactivation gates, each with a voltage dependent probability of being open (M and H). An important simplification in this model is that though several activation and inactivation gates can be connected in series in one channel, all act independently of each other and all obey the same voltage dependent equations. Transitions between the open and closed form of a gate follow the simple reaction: α M closed β M open Eq. 18 M closed = 1 M open Eq. 19 The rate factors α and β are voltage dependent. The equations for these rate factors are of the general form: α = b ( d + E ) c + e (d + E ) / f or α = a + be c + e (d +E )/f with a = d b Eq. 20 With a, b, c, d and f as terms which are specific for each conductance and E is the membrane potential. Singularities may be found in Hodgkin Huxley like equations when c=-1. If E is the opposite of d then the denominator becomes zero. Nodus obtains a value for α at this singular point by interpolation between values around the point, but the rate factor functions may still behave irregularly. Opening and closing of the gates can be described with the differential equation: M t = α( 1 M ) βm Eq. 21 From this equations follows the steady state open gate fraction M : M = α α +β Eq. 22 and the gate time constant τ M : 1 τ M = α +β Eq. 23
III-22 Modeling The ionic conductance itself can be described by one of the following equations: G (E,t ) = g Eq. 24 G (E,t ) = gm x Eq. 25 G (E,t ) = gm x H z Eq. 26 g-bar is the maximum conductance, which may be proportional to the membrane surface of the compartment (in the Nodus dialog windows and printouts it is marked as Gmax). Equation 24 describes a leak conductance, Eq. 25 and 26 are true voltage dependent ionic conductances with respectively x activation gates M, and x activation gates M and z inactivation gates H. Note that in Hodgkin-Huxley like equations inactivation is treated as another channel, but as inactivation increases the channel closes (go to zero) and vice versa. Equivalent circuit RI CM E Na G Na E K G K E L G L Figure III/7: equivalent circuit for a compartment with voltage dependent Na + and K + currents and a leak current. In compartments with excitable membrane the equivalent circuit is expanded, it includes for each voltage dependent current a variable conductance and a potential source (Fig. III/7). The circuit shown is for a compartment with the classic Hodgkin and Huxley currents. The ionic current I is determined by the conductance and its reversal potential E j : I k = G k ( E, t )( E E k ) Eq. 27 The first term in equation 4 is replaced by a term describing the voltage dependent ionic currents: E n t = j ( ) E n E j G j E n,t ( ) CM n + k E n E k RI nk CM n + I s CM n + I e CM n Eq. 28 For j ionic currents the voltage and time dependent conductance G j (E,t) (conductance is the inverse of resistance) is calculated and multiplied with the potential source which is determined by the difference between the membrane potential and the conductance specific reversal potential E j.
Modeling III-23 To model action potentials and other voltage dependent phenomena accurately, voltage clamp data for all ionic currents present in the neuron should be obtained to construct Hodgkin-Huxley equations. This represents a lot of work, and may not be experimentally feasible. Therefore a lot of authors use published equations. Try to measure the time constant, reversal potential and maximum conductance of the important ionic currents and adapt the standard equations, because these values vary between preparations. If these values cannot be measured directly with voltage clamps, find indirect measurements (refractory period, period of firing, etc.) and correct the values in the simulation by trial and error. Equations derived from voltage clamp experiments may not perform as expected in models where all the ionic currents were included. This is usually a consequence of the artificial conditions under which voltage clamp experiments were done (blockers, unusual ionic concentrations, etc.); in such cases G max may be changed. A small library of equations for voltage dependent ionic conductances is included on the Nodus Master disk. Nodus can plot any factor related to ionic conductances: α, β, M, H, τ M, τ H and G in factor versus membrane potential plots and M, H, τ M, τ N, G and I dynamically during simulations. These plots can help in developing new Hodgkin-Huxley like equations or in adapting existing equations to describe the conductances in the model. Simulation of Synapses and Connections Several functions to simulate synapses are available to the user. One can also implement special functions with user-defined processes (not available in Nodus 3.1). Presynaptic transmitter release In network models presynaptic activity fires synapses. Presynaptic transmitter release can be constant or variable. In both cases transmitter is released only when the presynaptic potential is higher than a threshold potential. With constant transmitter release the release is signaled to the postsynaptic cell once each time the threshold is crossed, where it arrives at a time t 0 which is needed to compute the postsynaptic conductance (use Eq. 33 or 34). With variable transmitter release the amount of transmitter is signaled continuously to the postsynaptic site (as long as the presynaptic potential is over the threshold); there is no t 0 available (use Eq. 32, 35 or 36). Three types of variable transmitter release T s are available: Linear to presynaptic potential: T s = b + f E E th ( ),T s b Eq. 29 The minimal transmitter release is the base amount b. The variable part is determined by the factor f and is computed relative to the threshold potential E th. Exponential to presynaptic potential: ( T s = b + fe E E th) /c,t s b Eq. 30 The minimal transmitter release is the base amount b. The variable part is determined by the factor f and the characteristic potential c; it is computed relative to the threshold potential E th. Process dependent T s = U( E,t, ),T s b Eq. 31 The transmitter release is determined by a user defined function U which is defined as a separate process.
III-24 Modeling Postsynaptic conductance Nodus has five types of postsynaptic conductances G s, which are grouped in 2 types. The type 1 synapses should be used if presynaptic transmitter release is constant or in single neuron models; type 2 synapses should be connected to variable transmitter presynaptic sites. Constant conductance (type 2) G s = gt s Eq. 32 g-bar is the maximum conductance (G max ), which may be proportional to the membrane surface of the compartment. The postsynaptic conductance is completely determined by the presynaptic variable transmitter release T s. Alpha function (type 1) G s = gt s t α e αt / τ Eq. 33 The postsynaptic conductance is determined by the alpha function, which is controlled by 2 factors: α and τ. The alpha function is computed relative to the start of the synaptic firing (t 0 = 0). Nodus uses a normalized version (i.e. 0<=G s <=1), it should be used with constant transmitter release. Factor α determines the steepness of the initial slope of the synaptic conductance, usually the value α=1 is used. The time to peak τ determines when the alpha function reaches its maximum conductance. Dual exponential function (type 1) ( ) G s = gt s e t / τ o e t / τ c Eq. 34 The postsynaptic conductance is determined by the dual exponential function, which is controlled by two time constants. The dual exponential function is computed relative to the start of the synaptic firing (t 0 = 0). Nodus uses a normalized version (i.e. 0<=G s <=1), it should be used with constant transmitter release. The open time constant τ ο determines how fast the conductance increases, while the close time constant τ c controls its decrease. Conductance dependent (type 2) G s = gt s G( E,t) Eq. 35 The postsynaptic conductance is voltage dependent and controlled by a Hodgkin-Huxley equation. Conductance dependent synapses should be used with variable transmitter release; the synapse can only be shut off by decreasing T s to zero. One can model NMDA-receptors with this type of postsynaptic conductance. Process dependent (type 2) G s = gt s U( E,t, ) Eq. 36 The postsynaptic conductance is determined by a user defined function U which is defined as a separate process. The postsynaptic current I s is given by: ( ) Eq. 37 I s =G s E E s E s is the synaptic reversal potential.
Modeling III-25 The electronic circuit for a passive compartment with one synaptic conductance is: RI CM E S G S E RM Figure III/8: equivalent circuit for a passive compartment with one synaptic conductance. Postsynaptic currents can be used in both network and single cell models in Nodus. In network models the presynaptic cell drives the synapse; in single cell models the user has to specify a synaptic firing time. To improve computation speed the user can specify a synaptic switch off time. This controls the switching off of alpha function or dual exponential function postsynaptic conductances. Without this switch off all synapses that have started firing would go on continuously, even when the actual conductance is infinitesimally small. The synapses will be switched off when the simulation time equals t 0 plus switch off time. Note that other types of synapses are always connected to variable presynaptic transmitter releases, they will be switched off when transmitter release goes to zero. Connections between neurons Connections between neurons in a network are defined in a straightforward manner. A presynaptic site is connected to a postsynaptic site and a delay time is specified. The delay time can be used to simulate axons; the presynaptic transmitter release will arrive at the postsynaptic site after the delay time has passed. A single presynaptic site can have connections with multiple postsynaptic sites (simulating multiple branchlets at the end of the axon). A single postsynaptic site can also have multiple presynaptic sites. One has to be careful about connecting the right type of transmitter release with the right synapse (type 1 or 2). Nodus does not check for this kind of error in connections and the result will be synapses that behave strangely (not firing at all, or firing infinitely). One should connect constant transmitter release to postsynaptic type 1 synapses (alpha functions or dual exponential functions) and variable transmitter release to type 2 synapses (constant, conductance or process dependent). Integration Methods Two integration methods are available: a fast hybrid Euler method for initial exploratory experiments and an accurate, slower fifth order Runge-Kutta method for confirmation of the final results, both with variable time steps. All simulation variables and integration steps are computed in double precision.
III-26 Modeling Both integration methods are sensitive to the stiffness of the differential equations, they will compute slowly if membrane voltage or concentration is changing very fast and/or if very small compartments are present in the model. The integration methods control the minor time step t. Nodus uses in simulations also a major time step, which is specified by the user. The major time step is used to control the switching on or off of experiments. The minor time step is always less than or equal to the major time step. Hybrid Euler method The hybrid method (Moore, J.W., and Ramon, F.: On numerical integration of the Hodgkin and Huxley equations for a membrane action potential. J. Theor. Biol., 45 (1974) 249-273) is a fast forward Euler method. (Eq. 38). The accuracy of the computation is controlled by setting absolute maxima for changes in membrane potential and concentration in any compartment. The minor time step is adapted dynamically to keep the changes below the maxima. E = E 0 + t E 0 t Eq. 38 For excitable membrane models the hybrid method simplifies the differential equations describing an excitable compartment (Eq. 20, 25 and 28) by making the rate factors for (in)activation (α and β) voltage independent for each time step t. The time constant τ M (Eq. 23) and the steady state open gate fraction M (Eq. 22) are determined from α and β. The approximate solution for M becomes then a simple exponential function: = M + ( M 0 M ) e t /τ M M Eq. 39 To increase computation speed a table of the (in)activation factors M and N is calculated before the simulation starts. An information dialog window shows the progress of this computation. The maximum error shown in the dialog window is estimated by halve the difference between the calculated value and the value obtained from interpolation between the preceding and the next entry in the table. If this error becomes too large one cannot trust the results from the hybrid Euler method. During the simulation the actual (in)activation factors are determined by linear interpolation between the values in the table. The hybrid method may oscillate when an equilibrium is reached in the simulation (for example after a current injection), more so in small compartments than in large ones. The error of the hybrid method relative to the Fehlberg method is usually smaller than 1%, input resistances and period of firing are less accurate than for example maximum amplitude of an action potential. Fehlberg method The Fehlberg method (Forsythe, G.E., Malcolm, M.A., and Moler, C.B.: Computer methods for mathematical computations. Prentice-Hall, Englewood Cliffs (1977) pp. 110-147) is based on the Runge-Kutta integration method. In its standard formulation the Runge-Kutta method is a fourth-order integration method: E = E 0 + k 1 + 2k 2 + 2k 3 +k 4 6 Eq. 40 k 1 = t E 0 t 0
Modeling III-27 k 2 = t E 0 + k 1 2 ( ) t 0 + t 2 E 0 + k 2 2 k 3 = t ( t 0 + t 2) ( ) t 0 + t ( ) E k 4 = t 0 +k 3 The Fehlberg method requires six function evaluations for each parameter. Five of these function values are combined to produce a fifth-order Runge-Kutta method, four values are used to produce a fourth-order method. Comparison of these two values yields a relative error estimate which is used to control the minor time step size. Parameters that are integrated include: membrane potential, concentrations, variable transmitter release, synaptic conductance and (in)activation factors for voltage dependent ionic conductances. The rate constants are integrated independent from the membrane potential (sequential mode of integration, Moore (1974)). The Fehlberg method is very accurate (minimum relative error can go down to 10-12 ), but it can be quite slow. It is sensitive to the stiffness of the differential equation, i.e. during sudden large changes in membrane potential (action potential, begin of a current injection, etc.) the minor time step can decrease enormously. The method may fail if the minor time step gets smaller than the minimum value which can be accurately represented as a double precision floating point number (0.11 10-15 ). In this case Nodus may temporarily increase the allowed relative error. An appropriate integration error message is shown; the change in relative error can be monitored with the Status Window command. Experiments Most experiments settings can change only when major time steps are reached. Some experimental parameters (like sinusoidal current injections) are however recomputed at every minor time step. Voltage clamp experiments are simulated by a post-integration correction. A normal integration step is performed in either of the integration methods. Then the clamping current is calculated and the change in membrane potential in the voltage clamped compartments is set to the difference with the clamping potential.
IV-28 Nodus Reference IV. NODUS REFERENCE Nodus Menus 2 7 The menu bar...27 Popup menus...28 Nodus Files 2 8 Overview...29 Links between files... 30 Subdefinitions in neuron definition file...33 Making backups of Nodus files...35 File compatibility with Nodus 2...35 Making New Models 3 6 Collect and organize the experimental data...36 Entering data into the definition files...37 Store the original model...40 Making New Simulations 4 1 What is a simulation database...41 Simulation database status and available menus...41 Selecting the model... 43 The initial value problem...44 Improving simulation speed...46 Selecting Simulation Parameters 4 7 The selection of a simulation parameter... 48 The compartment preselection popup menu...50 Error messages about parameter selection...51 Nodus Menus The menu bar Except for the File and Edit menus, all other menus in Nodus 3 are file specific (Simulation, Network, Neuron, Conductance). The file specific menu commands are enabled only when a window of the corresponding file type is active (i.e. is in the front) and the commands will operate on that file when invoked. Figure IV/1: the Nodus 3 menu bar. The File menu has a lot of submenus (marked with an arrow pointing to the right) to select on which type of file the command should work. To get the submenus press on the main menu command and then move to the submenu list while keeping the mouse button pressed, release the button over the required submenu command. Figure IV/2: the File menu with the Open submenus.
Nodus Reference IV-29 Several File menu commands can be modified by pressing the shift key and/or option key (see Appendix for a complete list). The names of the menu commands change as appropriate. Start pressing the shift and/or option key before selecting the menu! Any of the four simulation database windows (Plot Window, Time Window, Measure Window, Status Window) activates the Simulation menu. Some simulation menu commands may be enabled, but italicized (Fig. IV/18). This means that one can use the command to view settings, but that one cannot change the settings. At the bottom of the Network, Neuron and Conductance menus are lists of names of all the corresponding files loaded into memory (Fig. IV/3). Some of these files may have windows associated with them, they are shown with a check mark. Selecting one of the file names from these lists makes it the active window, by either bringing an existing window to the front or by making a new window. Neuron definition files in memory Figure IV/3: the Neuron menu. At the bottom of the menu is a list of 3 neuron definition files in memory. The Test-cell 7 definition window is active. Popup menus Nodus 3 makes extensive use of popup menus (Fig. IV/4): in most definition windows and dialog windows popup menus are present (see for example Fig. IV/9). Popup menus allow the user to make fast selections without having to remember names or index numbers. Figure IV/4: a popup menu shows the selected item (in this example the name of a conductance definition) when it is not pressed (left). When the user presses the menu a list of all available menu items is shown (right). Popup menus require that all definitions and subdefinitions in Nodus are named; naming neuron compartments can also help in later popup selections. Try to use self-explanatory names, it makes no sense to have a popup menu full of incomprehensible acronyms or numbers. Do not use very long names; they may not fit in the dialog windows so that only the first part of the name will be shown. Nodus Files Nodus 3 can make seven different types of files and some files may be linked to other files. The complete user interface is based upon the Nodus file structure, so it is quite important to understand what each file type represents and how the links between files work.
IV-30 Nodus Reference Overview Two file types used by Nodus 3 cannot be accessed by the user. Nodus Preferences file: The Nodus Preferences file contains the default and user specified settings for the Nodus application. Nodus needs the Nodus Preferences file to run, it is personalized for each registered user. It cannot make a default preference file in case of disk failure, so be sure to always have a backup of this file available. The Nodus Preferences file should be placed together with the Nodus application file (i.e. both on the desktop or both in the same folder) or it can be placed in the System Folder. Nodus Resume file: The Nodus Resume file is created when the user quits Nodus while a simulation is running and the Automatic Saving command is selected. It contains the data necessary to continue the simulation after restarting Nodus. Nodus creates and deletes this file when appropriate, it is always put at the same desktop level as the Nodus application. Do not rename this file. If it is renamed or deleted Nodus will not be able to continue the simulation after restarting. Four file types are accessible to the user; the File menu commands New, Open, Close and Save operate on these files. They contain simulation databases and model definitions. Simulation data file: A simulation data file contains all the data necessary to run a simulation. This includes the compiled simulation database, initial values, settings for the integration method, for graphic and text output and for the experiments, and the links to the original model definition files. It does not contain any simulation results. Network definition file: A network definition file contains the description of a small network model. This includes the local names of all the neurons, links to the original neuron definition files and data about the synaptic connections between the neurons. Neuron definition file: A neuron definition file contains the complete description of a compartmental neuron model. This includes the cable parameters and morphology of all the compartments and the topography of the connections between compartments. The file also contains the subdefinitions for all the ionic currents, synaptic currents or transmitter release sites used in the neuron model. Ionic current subdefinitions and some synaptic current subdefinitions have links to the equations in conductance definition files. Conductance definition file: A conductance definition file contains the equations in Hodgkin-Huxley like format for a conductance. Nodus can make one type of output file. It can only be opened from other applications. Text output file: A text output file contains the results of a simulation in ASCI format. Text is separated by spaces or TABs, lines are separated by carriage returns. Text output files can be opened as a standard Macintosh text file by any word processing software package or they can be imported into a spreadsheet, graphics or statistical packages for display or further analysis of the data.
Nodus Reference IV-31 Links between files The four user accessible files are hierarchically ordered by the links possible between them (Fig. IV/5). The advantage of this structure is that information can be spread over different files. Common data has to be defined only once in a file at a lower level of the hierarchy; it can then be used by different models higher up in the file hierarchy. Files Links Max number in memory Simulation Data File Simulation database Integration settings Plot and text output settings Experiment controls 1 Network Definition File Local neuron names Connections OR 1 Neuron Definition File Compartments Ionic Current subdefinitions Synaptic Current subdefinitions Transmitter Release subdefinitions 20 Conductance Definit. File Equations 20 Figure IV/5: the Nodus file hierarchy and the links possible between files. Links between files are always stored and defined in the files at the top of the hierarchy (Fig. IV/6). Links work best if all linked files are in the same folder. The user makes a link by selecting the name of the lower level file with a popup menu in a dialog window belonging to the higher level file (Fig. IV/4). Links are just like any other piece of data; changes to file links are permanent only after they have been Saved to the disk file. Nodus finds linked files by their file name, so the user should never change file names (at the Finder level) because then the file will no longer be recognized. All definition files also have a hidden file ID-number which must match the one stored as a link in the file at a higher level in the hierarchy. The hidden ID-numbers are unique for each file so that all links are always unique. The hidden ID-number changes when a Save As command is done (Figs. IV/6 & 7). At that time the user can change all links in higher level files to point to the new file (with new name and ID-number) by selecting the Update all links in memory check box in the dialog window (Fig. IV/6, bottom right); otherwise the files at higher levels in the hierarchy will continue to point at the original file (Fig. IV/6, bottom center). Note that at present only links in files loaded into memory can be updated, disk files are not changed.
IV-32 On disk Neuron A ID #94532 20 compartments Link to 'Funny', #4589 Subdefinitions, links Neuron B ID #3738 568 compartments Link to 'Funny', #4589 Subdefinitions, links Funny ID #4589 am = 0.01 / exp( ) In memory Nodus Reference Open Neuron 'A' Neuron A ID #94532 20 compartments a link to #4589 Subdefinitions, other links Funny ID #4589 am = 0.01 / exp((40+e)/80) User edits 'Funny' Neuron A ID #94532 20 compartments a link to #4589 Subdefinitions, other links Save 'Funny' Neuron A ID #94532 20 compartments Link to 'Funny', #4589 Subdefinitions, links Neuron B ID #3738 568 compartments Link to 'Funny', #4589 Subdefinitions, links Funny ID #4589 am = 0.05/ exp( ) On disk Save As 'Fast Funny' without Update all links Neuron A ID #94532 20 compartments Link to 'Funny', #4589 Subdefinitions, links Neuron B ID #3738 568 compartments Link to 'Funny', #4589 Subdefinitions, links Funny ID #4589 am = 0.01 / exp( ) Fast Funny ID #235498 am = 0.05 / exp( ) Funny ID #4589 am = 0.05 / exp((40+e)/80) Save As 'Fast Funny' with Update all links Neuron A ID #94532 20 compartments Link to 'Fast Funny', #53849 Neuron B ID #3738 568 compartments Link to 'Funny', #4589 Subdefinitions, links Funny ID #4589 am = 0.01 / exp( ) Fast Funny ID #53849 am = 0.05 / exp( ) Figure IV/6: links between files before and after a Save or Save As command. The ID# numbers are invisible to the user. Refer to the text for an explanation.
Nodus Reference IV-33 Figure IV/7: the Save As file dialog window. At the bottom of the hierarchy are conductance definition files which contain the equations for ionic conductances. They do not have links, but are a good example of the usefulness of the linked files concept. Several neuron models may use the same set of ionic conducance equations, for example the standard Hodgkin-Huxley equations for the fast sodium and delayed rectifier currents. These equations have to be defined only once in 2 conductance definition files (the files HH Fast Na Current and HH Delayed Rectifier on the Nodus Master Disk), which can then be linked to models of the Squid Giant Axon and to any number of other models. Neuron definition files can have links to conductance definition files in subdefinitions. Ionic currents are defined by specifying a reversal potential and a maximum conductance, which are saved in the neuron definition file, and a link to a set of conductance equations in a conductance definition file. Synaptic currents may also use conductance equations from a conductance definition file. A neuron definition file can be viewed and edited while a linked conductance definition file is not in memory, except for the affected ionic and synaptic current subdefinitions. Network definition files always have links to neuron definition files. A network can consist of several instances of one neuron model (like in the Test 7 network on the Nodus Master Disk), or can combine different neuron models. A network definition file cannot be used if the linked neuron definition files are not in memory. Simulation data files have links to all model definition files which were used as sources to compile the simulation database. A simulation data file knows only the source model definition file selected in the New Simulation dialog by name; for the lower level files (for example conductance definition files) only hidden file ID-numbers are available. When a file is Opened all lower level linked files in the same folder are also loaded into memory, including files connected by secondary links. For example when the Test 7 network network definition file is Opened the linked neuron definition file Test-cell 7 is loaded and 3 conductance definition files which are linked to Test-cell 7 are also loaded. No windows are created for automatically loaded linked files, but their names are listed at the bottom of the corresponding menus. Select the name in the menu to make a file window. Figure IV/8: the Open file dialog window.
IV-34 Nodus Reference Note again that successful loading of linked files depends on their file name, which should not have changed. When a file cannot be found, an extra Open dialog window prompts the user to open it manually. When the top level file is then Saved, the new file name for the linked file will be stored. Automatic loading of linked files can be prevented by switching off the Open all linked files check box in the Open dialog window. Changing data in lower level files will affect all linked model definition files higher up. This can be dangerous if the lower level file is linked to a lot of different model definition files. For example if the original conductance definition files of the standard Hodgkin-Huxley equations are changed to slow the rate factors down for an invertebrate neuron model; not only the invertebrate neuron model will be affected, but also the Squid Giant Axon model! The solution is to give the changed conductance definition files a new name using the Save As command with the Update all links in memory check box switched on (while the invertebrate neuron model is in memory and the Squid Giant Axon model is not). The Squid Giant Axon and the invertebrate neuron model (do not forget to Save it!) will then be linked to different conductance definition files. See Fig. IV/6 for additional information. Simulation data files are not affected by changes in their model files, though some changes may interfere with selecting parameters with popup menus (see the section on Selecting Simulation Parameters ). Changes in model definition files affect the simulation data file when a New Simulation database is compiled. Subdefinitions in neuron definition files Neuron definition files can contain subdefinitions for Ionic Currents, Synaptic Currents or Transmitter Release sites. The use of subdefinitions makes the creation and management of these features very flexible. The subdefinitions contain all the data necessary to implement the feature (see further) except for the location within the compartmental neuron model. At individual compartments the user can then select one or several of the subdefinitions to tie them to the compartment (Fig IV/9). If no subdefinitions are selected at a compartment it is a passive compartment without pre- or postsynaptic contacts. If a set of ionic currents is selected it becomes an excitable compartment, if synaptic currents are selected it is a postsynaptic compartment or if a transmitter release site is selected a presynaptic compartment. Several subdefinitions can be combined, e.g. to make an excitable postsynaptic compartment (Fig IV/9), and any combination of different types of compartments is allowed within a neuron model. The Squid Giant Axon model contains only excitable compartments, all using the same Hodgkin-Huxley Currents subdefinition; the Testcell 6 model is completely passive, but has different postsynaptic currents in all its compartments; the Test-cell 7 model has passive, excitable and postsynaptic compartments. Figure IV/9: ionic current subdefinition selection in a compartment definition window. This compartment also has one postsynaptic site, but no presynaptic site.
Nodus Reference IV-35 Subdefinitions can be created, viewed and edited separately from the neuron model(s) that use them by selecting the corresponding menu commands in the Neuron menu. As an example the Synaptic Currents subdefinition dialog is shown (Fig. IV/10), the other subdefinition dialog windows are very similar. At the left side of the dialog window are the controls and data specific for a selected subdefinition. At the top left of the subdefinition dialog window is a subdefinition selection popup menu labeled as Synapse subdefinition, it contains a list of the names of all the synaptic current subdefinitions in memory (from all the neuron definition files in memory). If the neuron definition file in the front window uses synaptic current subdefinitions, the name of the first one used will be shown (as in Fig. IV/10: Slow EPSP); if no synaptic current subdefinitions are in memory None will be shown. Under the subdefinition selection popup menu the data box shows parameters specific for the subdefinition; in this case a peak conductance, a reversal potential and a variable synaptic conductance. If no subdefinition is selected, the data box is empty and all its options are dimmed. Figure IV/10: the Synaptic Currents subdefinition dialog window. At the right side of the dialog window are general subdefinition controls and info. At the top is a list of all neuron definitions in memory using the selected subdefinition. Below it are four buttons that perform subdefinition management: New, Delete, Duplicate and Rename. Their function is implicit in their names. The Delete, Duplicate and Rename buttons act on the presently shown subdefinition; the New button creates a new, empty subdefinition with the name Untitled. At the bottom right side are the dialog window controls, the OK and Cancel buttons. Both close the dialog window. Pressing OK keeps any changes made to the data box entries, pressing Cancel discards all changes made to the currently selected subdefinition. The subdefinition selection popup menu makes a rapid selection of a subdefinition for viewing or editing possible. Note however that changing the selected subdefinition is equivalent to pressing OK without closing the dialog window, any changes to the subdefinition shown will be preserved. An Ionic Currents subdefinition is a set of up to ten ionic currents. For each current it contains the maximum conductance (either as an absolute value in ns or as a compartment membrane surface dependent value in ms/cm 2 ), the reversal potential and a link to the voltage dependent conductance equations in a conductance definition file (except for the Leak, which has a constant conductance). A Synaptic Currents subdefinition specifies one postsynaptic site. Up to three different subdefinitions can be used in a single compartment. Though Nodus does not prevent the repeated use of the same subdefinition at a single compartment, this is not useful (in network models several presynaptic cells can be connected to the same postsynaptic site, i.e. to the same postsynaptic subdefinition in a compartment). The subdefinition contains the synaptic peak conductance, the reversal potential and variable conductance data.
IV-36 Nodus Reference The synaptic conductance can be constant; this is useful for synapses with graded transmitter release where the user wants to specify the change in synaptic conductance at the presynaptic site. Variable synaptic conductances can be an alpha or dual exponential function (see chapter III), a voltage dependent conductance defined by a link to a conductance definition file or an user defined process. A Transmitter Release subdefinition specifies one presynaptic site. A compartment can contain only one presynaptic site. The subdefinition contains the minimal amount of transmitter released and a threshold membrane potential above which transmitter is released. Variable (graded) transmitter release is controlled by linear or exponential voltage dependent functions or by a user defined process. Subdefinitions are stored in the neuron definition files. When a neuron model is Saved all subdefinitions used by that model are stored in the neuron definition file. When a neuron definition file is Opened all subdefinitions present in the file are loaded into memory and added to the lists shown in the respective selection popup menus. Of each type of subdefinition up to 20 different ones can be stored in memory. If a subdefinition with the same name is already in memory, Nodus will add #0 to the name (or a consecutive number if the subdefinition name already contained a #number ) unless Multiple use of subdefinitions is allowed (Preferences command, see next paragraph). Subdefinitions are very flexible. Global changes which affect several or all compartments are easy to make, while the user also has full control over the features used in any single compartment. There is one potentially dangerous feature: the use of the same subdefinition in several (different) neuron definition files. If any changes are made to such a subdefinition, it will only be saved (when the user Saves the file) to those neuron definition files which were in memory at that time; the other files will still contain the old version of the subdefinition! This is usually undesirable and therefore subdefinitions can be used in different neuron definition files only when the Multiple use of subdefinitions check box in the Preferences dialog is switched on (the default setting is off). Multiple use of subdefinitions can be useful in network models, where for example a type of postsynaptic site might be common to different neuron models. If the Multiple use of subdefinitions is enabled, then always Open the Network (this guarantees that all the required neuron definition files are in memory) instead of opening individual neuron definition files. Making backups of Nodus files One should always make backups of important files to protect oneself against hard disk failures or viral infections. This rule applies also to Nodus files, but one has to take into account the specific properties of the Nodus file hierarchy. - Make your backup copies always at the Finder level. Do not use Duplicate because that Finder command changes the file names. Drag the file icons to the backup disk, the Finder will copy the files and keep the original files. If you want to store backups in another folder on the same disk as the original files, press the option key while dragging the file icons to the folder (they will be copied instead of moved). - Never use the Save As command to make backup files. This command changes the hidden ID-number of the file so that it will no longer be recognized by linked files! - Always make backups of all linked Nodus files together. They need each other to work properly. If you need to restore damaged files from backups, it is usually best to restore all linked files together also. File compatibility with Nodus 2 Nodus 2.0-2.3 neuron and conductance definition files can be read by Nodus 3, but they are not linked together till they are saved as Nodus 3 files. You should open the Nodus 2 conductance definition files before the neuron definition files.
Nodus Reference IV-37 Synaptic conductance is computed by different equations in Nodus 3; synaptic conductance settings of Nodus 2 files are converted to approximate the old conductances in Nodus 3 but the results will not be identical. The subdefinition concept was not available in Nodus 2, were these settings had to be repeated for each compartment. Nodus 3 tries to collect all ionic current or synaptic current settings from one neuron definition into one or a few subdefinitions, but success is not guaranteed. If you have a large set of Nodus 2 neuron and conductance definition files you should convert all of them to Nodus 3 format at once. Do not mix the old Nodus 2 and converted or new Nodus 3 files in the same folder; it might confuse Nodus if they have the same names. Nodus 2.0-2.3 simulation data files and simulation run files cannot be read by Nodus 3. The internal format of the simulation databases has completely changed and simulation plot, output and experiment settings are managed by new, more user friendly methods. All simulation database computations in Nodus 3 are done in double precision. Because the compartment capacitance and the linking resistances are computed in double precision high accuracy simulations of passive neuron models in Nodus 3 may produce slightly different results than in Nodus 2 (less than 0.1% for R N and τ m in my experience). Nodus 3 is always more accurate (in Nodus 2 compartment capacitance and the linking resistances were computed in single precision and then converted to double precision if double precision integration was requested). Making New Models Theoretical aspects of modeling with Nodus are covered in chapter III. This section offers practical suggestions about constructing and managing your own models. The systematic approach to modeling presented here applies to doing real work with Nodus. To get acquainted with Nodus play around with the example files and try to make small models. Collect and organize the experimental data One can model to test and explore theoretical hypotheses, or one can model an experimental preparation to simulate experiments and test whether scientific knowledge about it is complete. In both cases the modeler has to supply several parameters to Nodus, which usually are derived from experimental measurements. Decide on what type of model is going to be used. Will it have passive or excitable membrane? In a passive membrane model, will R m be constant or variable? Can the synaptic events be examined with a single neuron model or is a simulation of the presynaptic neuron(s) with a small network model necessary? Once such decisions have been made, make a list of all the (sub)- definitions that are going to be used: the number of ionic currents, the number of neurons, different types of synapses, etc. For the (sub)definitions the following parameters are needed: Ionic currents: (approximate) equations in Hodgkin-Huxley format. Reversal potential, maximum conductance (may be distinct for different neurons or compartments). Postsynaptic sites: (approximate) equations for the variable synaptic conductance. Reversal potential, maximum conductance (may be distinct for different neurons or compartments). Presynaptic site: constant or graded (variable) transmitter release. Threshold potential for transmitter release and minimum amount released (may be distinct for different neurons or compartments). For graded transmitter release the equations describing the amount released. Neurons: morphological data (number of compartments, sizes of compartments, connections). The cable parameters: C m, R m, R i. Resting membrane potential. Distribution and location of ionic currents, pre- and postsynaptic sites. Networks: number and types of neurons. Connections between neurons: presynaptic to postsynaptic site types, delay times (to simulate axon length).
IV-38 Nodus Reference Start with collecting all the parameters listed and try to put them in a format that approaches the way the information is structured in Nodus. Usually some parameters are missing. Maybe something can be found in old lab books or in the literature. If no experimental data are available one has to use default values, usually based on experimental data from other preparations. A good example is the specific capacitance; the default value of 1 µf/cm 2 is almost never checked experimentally. When in doubt, consult the modeling literature to see how colleagues solved similar problems. Often an important aspect of modeling is to examine how different values for an unknown parameter influence the ability of the model to simulate certain experiments. Nodus makes it easy to change model parameters, so it is not a tragedy if some of the initial parameters are found to be unsuitable during actual simulations. While a lot of experimental data may need extensive work to get it into formats compatible with Nodus (for example getting conductance equations from voltage clamp data), morphological data can often be used without any changes. The user can optimize the neuron model in Nodus either manually (Fuse Compartments, Split Compartment) or automatically (Optimize Model) to get acceptable performance during simulations. When the neuron models are large, i.e. they have a lot of compartments, one should examine whether they can be imported into Nodus. Detailed morphological data are usually stored in computer files. Transferring these files from other computer systems to the Macintosh is easy, use the file exchange software supplied by Apple. Compare the format of the morphological data with the formats available in the Import Neuron command (see Appendix). If the differences are small, try to convert the files yourself. A spreadsheet is usually the best platform for editing morphological data files, most packages can read text files. Do not forget to save the file in text format (a golden tip: Microsoft Excel does not put a carriage return after the last row of a spreadsheet; make a fake row at the end of the spreadsheet by putting a non-numeric character in its first cell). If the format of the (large) morphological data file is quite different from the ones available in Nodus you may consult the author and request the addition of a new import format. In the mean time organize the data so that it is easy to type them into Nodus manually. Give each compartment a compartment number and list its diameter, length, 3-dimensional coordinates (if available) and all connections to other compartments. The soma should be the first compartment (#1), number other compartments from proximal to distal in either of two ways. One can first number the large processes (number the compartments of the stem of the first dendrite from 2 to n, the stem of the second dendrite from n+1 onward, etc) and then number the smaller dendritic branches. Or all compartments (stem and all branches) of one dendrite can be numbered first and then the next dendrite (or the axon) can be numbered, etc. The first numbering scheme can have advantages when variable R m is used, the second one is more intuitive and corresponds to the format of most morphology files. Entering data into the definition files Links to files lower in the hierarchy can only be established after these files have been created, therefore one should start by making the files at the bottom of the file hierarchy. All definition file windows have a text box that initially contains the word Comment (Fig. IV/11). Use this space to identify the definition by either quoting references to the original data or by a note specifying what makes this particular definition file different from similar ones. Conductance definition files: First make the conductance definition files with New Conductance and type the equation parameters into the conductance definition window (Fig. IV/11). Push M to define the activation factor equations or H to define the inactivation factor equations. Save the files with recognizable names.
Nodus Reference IV-39 If the conductance equations used are approximate, they may need to be changed later. Consider then a way to mark in the file name the progression of changes to the equations, usually a numbering scheme is appropriate. Figure IV/11: the empty conductance definition window. Check whether the equations behave as predicted. Plot (In)Activation factors and Time Constants over the relevant membrane potential range. This is a quick way to find typing errors. If the equations were never tested in simulations it might be worth checking them in a small test model with only one compartment (this will be much quicker). See whether they replicate voltage clamp experiments. Neuron definition files : Start with the subdefinitions first. Define all Ionic Currents, Synaptic Currents and Transmitter Release subdefinitions needed in the model. Again, consider that usually a lot of changes need to be made to subdefinitions during the tuning of the model (particularly the maximum/peak conductances are often changed a lot). Decide which compartment labels are going to be used for the selection of simulation parameters (refer to the end of this chapter). One can either name all compartments or only the interesting ones (like the soma, dendritic roots, synaptic sites, etc). Another labeling method is to use the Structure type popup menu (Fig. IV/13) to describe the structure of each compartment. Both the Names and Structure type of a compartment are optional. They are worth using in large neuron models, because compartment numbers are not very intuitive and may change after an Optimize Model command. Good planning and consistency of compartment labels can make using Nodus more rewarding. Figure IV/12: the default neuron definition window. Make the compartmental model of the neuron. If the compartmental model can be imported do Import Neuron (and life is easy) otherwise do New Neuron. Type in the Cable parameters for the neuron (Fig. IV/12).
IV-40 Nodus Reference Decide on whether the Tree format and 3-dim coordinates options are going to be used. The rules imposed by the Tree format option are discussed in chapter III. Always use the Tree format option, unless it interferes in some way with the neuron model itself. Using 3- dim coordinates has no advantages in Nodus 3.1, it reduces the maximum number of compartments from 4000 to 3000. Future Nodus versions may have 3-dim drawings, which will need these coordinates. Save the neuron definition file with a recognizable name. If the neuron was made with the New Neuron command change the Number of compartments into the correct number. Then do Next Compartment to start defining compartment sizes and connections (Fig. IV/13). Most of this work can be done from the keyboard by tabbing from one text entry box to the next and pressing command-+ to go to the next compartment. Note that Nodus automatically connects distal compartments to their proximal parents (if the Tree format option is switched on); these back connections are disabled and cannot be edited. Do not use branch connections or weight factors (explained in chapter III) unless you understand their (dis)advantages. Nodus checks whether all the entered values are acceptable before allowing the user to go to another compartment. Save the neuron definition file frequently. Print the complete neuron definition file and check for typing errors in compartment sizes or connections. Figure IV/13: an empty compartment definition window. Compartment Names and Structure types are also entered in the compartment definition window (Fig. IV/13). If New Neuron was used this can be done during the entry of compartment sizes and connections; if Import Neuron was used this can be done with consecutive Next Compartments to label each compartment or Go to Compartment # to label only interesting ones. At the same time subdefinitions can be tied to the compartment. Save the neuron definition file frequently. Check whether the neuron model behaves as expected in simulations of experiments. Perform tuning of the cable parameters, the maximum/peak conductances, etc as needed. If the simulations go too slow (with large models) one might make special versions of the neuron definition files for different experiments. One trick is too eliminate all subdefinitions that are not used in a specific simulation. For example to test input resistance and peel exponentials of a large passive membrane model one does not need to have the synaptic current subdefinitions. If a lot of postsynaptic sites are defined on the model eliminating all these subdefinitions with the Delete button in the Synaptic Currents dialog window will create a model specific for the peeling experiment that computes much faster. One might also consider simplifying the morphological complexity of the model with the Optimize Model command to reduce the number of compartments during the tuning of ionic currents subdefinitions.
Nodus Reference IV-41 Network definition file: Finally the network definition file is made. Be sure to have all the needed neuron definition files in memory. Do New Network and enter all the network neurons (Fig. IV/14). Press the Neuron definition popup menus in the middle column to add a neuron model to the network and give it an appropriate Local name. Local names will usually be either specific names (like Pyloric dilator ) or reflect anatomical location (left, right, dorsal, ventral, etc.); they will be used in the selection popup menus. Press the Next button to enter more than 8 neuron models. Figure IV/14: the empty network definition window. Save the network definition file with a recognizable name. Specify the connections between the neurons with the Set Connections command, the network definition window will show the total number of pre- (Out) en postsynaptic (In) connections. Save the file again. Store the original models A final simulation model ready for publication will usually be quite different from the initial model entered into Nodus. The initial model however contained unmodified experimental data and is as such a good condensation of the available biological data. It is worth keeping it as a summary for later reference. The same goes for intermediary models created during the tuning process and the progressive series of simulations. Instead of printing all the models out and pasting long listings in the lab book, one can just refer to the model file names in the lab book and note down simulation results while storing the Nodus files. Again, clear names (with a numbering scheme) and sensible use of the Comment space will help to identify them later on. Remember to store the linked lower level files also. Storing original neuron definition files is important if the commands Fuse Compartments, Split Compartment or Optimize Model are used. These commands are very useful during the tuning process, because changes to the cable parameters may alter electrotonic lengths enough to make compartments too long or too short. But one should not repeat these commands too often on the same file, because the neuron model morphology will slowly divert from the original measured data. It is better to go back to a neuron definition file with the original morphology, enter all the changes to the cable parameters and Optimize Model again.
IV-42 Nodus Reference Making New Simulations What is a simulation database Model definition files are a good framework to define and manage neuron or network models because they were structured to look like standard physiological and morphological data. They have however an inefficient format to run simulations with. Nodus compiles all data from a selected model definition file (and from the linked files) into one simulation database. The simulation database structure was optimized for simulation speed; unfortunately this resulted in a incomprehensible internal format. The format of the database is invisible to the user and of no concern as all relevant simulation parameters can be accessed with selection popup menus (see the next section). There can be only one simulation database in memory. The simulation databases are numbered consecutively. The simulation number is shown in the simulation plot window title as.sim0001. The simulation number counter can be reset to zero in the Preferences dialog window. The simulation database is a separate entity. The model parameters in a simulation database do not change when the original model definition files are changed (the user can change some parameters in the database with the View/Edit Parameters command). A New Simulation database has to be compiled to convey changes from the model files. Editing of model files is possible while a simulation is running. The parameter selection popup menus in simulation commands depend however on the content of the source model definition files; they may not work as expected if substantial changes were made to the model. Simulation database status and available menus The status of the simulation database determines which Simulation menu commands are enabled. Four alternatives are possible: - No simulation database is in memory or the preceding simulation was Closed. All the Simulation menu commands are dimmed. Open Simulation or do New Simulation to get a simulation database. Integration commands Output commands Experiment commands Window commands Figure IV/15: the Simulation menu when there is no simulation database in memory or the preceding simulation was Closed. - A fresh simulation database is in memory. This is a new simulation which has never Run. All Simulation menu commands are available. Experiment commands and the Text Output command are marked if they are active. The Plot Window command can be used to make it the frontmost window if necessary. The integration, output and experiment commands are used to prepare the simulation. The Configure Plots command works only on fresh simulation databases.
Nodus Reference IV-43 Integration commands Output commands Experiment commands Window commands Figure IV/16: the Simulation menu when there is a fresh simulation database in memory. Text Output and a Current Clamp experiment are active, the Plot Window and Time Window are shown. - A started simulation database is in memory This simulation is Running (or has been running but the user Paused). All Simulation menu commands are available. The Run command changes to Pause while the simulation is running. The Configure Plot command is italicized, Text Output is italicized if in use. The integration commands work only during Pauses. All the experiment commands work, but be cautious with making changes to experiments while the simulation is running. Most changes to the experiment settings will take effect from the next major time step on; retroactive changes are not possible. Integration commands Output commands Experiment commands Window commands Figure IV/17: the Simulation menu while a simulation is Running. Text Output and a Current Clamp experiment are active, the Plot Window and Time Window are shown. - An old simulation database is in memory The simulation has ended (the simulation time equals the End time in the Integration Settings dialog). The window commands (Measure included) work. Other Simulation menu commands are available if they were active (marked) during the simulation, but italicized to show that one can use the commands only to view parameters and settings. Menu commands that were not used are dimmed. The title of the plot window is added to the list at the bottom of the Simulation menu. Because only one simulation database can be in memory one has to Close the simulation before a New or Open Simulation command is possible. This can be done in two ways: Close simulation window title (default) closes the simulation database and all its windows; Close simulation (press the option key while selecting the menu) closes the simulation database but keeps the plot window (it is listed at the bottom of the Simulation menu).
IV-44 Nodus Reference Integration commands Output commands Experiment commands Window commands Old simulation windows Figure IV/18: the Simulation menu when there is an old simulation database in memory. The simulation has not been closed yet, it had text output and a current clamp experiment. Close simulation dims all Simulation menu commands and one can no longer access the results or parameters (except for the graphic output if the plot window was not closed). The old simulation database remains however in memory and it can be the template for the next simulation database. Selecting the model for a simulation database The New Simulation command makes a new simulation database (Fig. IV/19), with a simulation number one higher than the preceding one. The content of the new simulation database can come from one of three sources. If an old database is in memory it can be used as a template for the new one by selecting the Use option. This is fast because it requires no database compilation. The new simulation database is identical to the preceding one. Use this option to run several experiments on the same model or to repeat a simulation with other integration or output settings. Figure IV/19: the New Simulation command dialog window. If no old simulation database is available, or to run a simulation of a different model or of a slightly changed model, select a the Compile from options with either a network definition or neuron definition file as the source from which a new simulation database will be compiled. The available source files presented by the New Simulation dialog (Fig. IV/19) depend on what is in memory. Simulations can be compiled only if the source model file and all linked files are loaded. Any changes to the source model definition files will be included in the simulation database, even if they have not been Saved yet.
Nodus Reference IV-45 Compiling a new simulation database may take several seconds, depending on the number of compartments and subdefinitions in the model. After the compilation is finished the new simulation database will conform completely to all the source definition files. The newly compiled database will inherit its integration and graphic output settings from the preceding one, or it will have default settings if no simulation database was in memory. Inheritance of experiment and text output settings is under user control. If the Clear experiment & text output check box is switched on (Fig. IV/19) all these settings will return to their default values. The inheritance of simulation settings can be a powerful tool to create a simulation of a new model without having to specify all the settings. If the correct settings are in memory (for example from a simulation data file after an Open Simulation) they will work on the new model also. If the new model is quite different from the preceding one, Nodus tries to apply the settings to the new model using comparison techniques. The comparisons are based on the order in which (sub)definitions were used and on hidden ID-numbers for compartments. When editing model files one should try not to change the order of the neurons in a network definition file, of the conductances in a Ionic Currents subdefinition or of the compartment ties to Synaptic Currents subdefinitions. The advantage of this approach is that when any of these (sub)definitions is replaced in a model by an analogue (for example a better conductance equation), all simulation parameters settings will remain operational. A disadvantage is that when a (sub)definitions is deleted, all settings referring to it or to subsequent (sub)definitions will seem to shift. If Nodus does not succeed in applying the settings an alert listing all the problems will warn the user, use the appropriate menu commands to correct these settings. The initial value problem To Run a simulation one needs initial values for all the parameters to start the computations from. All the compartments in the neuron(s) have an initial membrane potential and if subdefinitions are used several other parameters also need initial values. The user has limited global control over initial values and full single value control. After Open Simulation the initial values are loaded from the simulation definition file; these values are the parameter values present in the simulation database when it was Saved. The New Simulation dialog has several options for automatic initial value generation (Fig. IV/19). Initial values can be extremely important to the success of a simulation, especially if voltage dependent ionic conductances are used in the model. If conductance (in)activation factors are not matched initially the simulated neuron may behave irregularly (period of firing will be variable, etc.) till a dynamic equilibrium is reached. In some cases the simulation may reach an equilibrium which has no biological meaning (for example a continuous depolarization instead of firing), caused by an initial imbalance in conductance (in)activation factors which was too large to get corrected spontaneously. The New Simulation dialog window (Fig. IV/19) presents up to three options for automatic generation of initial values. One of these options will always be selected as default, this can be controlled with the Preferences menu. The Values in memory option is available only if an old simulation database is in memory. Figure IV/20: error message during execution of a New Simulation command with Values in memory option for the Initial values. For some simulation parameters no old values were available, they will be initialized with the Resting potential option.
IV-46 Nodus Reference For complex models Values in memory is often the best option, in this way one may start with an equilibrated simulation. The initial values can be loaded from a simulation data file with Open Simulation; if one wants to simulate a slightly different model Close it again and do a New Simulation of the desired model. If the new model is quite different from the preceding one, Nodus tries to find initial values for all the model parameters using comparison techniques. If Nodus cannot find some parameters in the old simulation database, it alerts the user and uses the Resting potential option for these parameters (Fig. IV/20). mv 20 0-20 -40-60 1.0 0.8 0.6 0.4 0.2 0.0 1.0 0.8 0.6 0.4 0.2 0.0 0 100 200 300 400 500 ms Figure IV/20: the first 600 ms from the Test-cell 2 Demo simulation. Upper axis: membrane potential in the second compartment (were the ionic currents are located). Middle axis: conductance activation (fat lines) and inactivation (thin lines) for the CS Fast Na Current ; full lines: dynamic values computed by the simulation; broken lines: static equilibrium values (M and N ) for the membrane potential at that time. Lower axis: same for the CS Delayed Rectifier.
Nodus Reference IV-47 The Resting potential and Set to mv work in similar ways; in the first case all compartments are initialized to their specific resting membrane potential, otherwise they are all set to the same (user specified) potential. All the other parameters are set at theoretical equilibrium values for the selected membrane potential. Note that this is a static equilibrium as opposed to a dynamic equilibrium that the simulation can settle in. Consider the firing neuron from the equilibrated Test-cell 2 Demo simulation (Fig. IV/21). The conductance (in)activation factors are changing in a smooth fashion and the membrane voltage is also changing continuously, as a result the (in)activation factors never reach their theoretical equilibrium value (M and N ) though they approach it during the slow repolarization phase. The combination of all these parameters at a precise time in the simulation constitute a dynamic equilibrium, which usually cannot be reconstructed analytically. Initial values may also be changed manually with the View/Edit Parameters command after the simulation database has been compiled. This is not recommended for the inexperienced user as one can easily introduce combinations of parameters that produce unexpected results. Note also that some parameters are stored in a derived format, see the description in the section on the Simulation menu in chapter V. In network models on going synaptic events may be considered initial values also. Synaptic firing times can be set explicitly at the postsynaptic sites with the corresponding command, but the synapses cannot be set to have begun firing before the start of the simulation. The only way to have on going synaptic events when the simulation starts is to have old synaptic events from a preceding simulation present. Usually one wants to Clear all synaptic events from preceding simulations. This is the default option presented in the New Simulation and Open Simulation dialog windows (Fig. IV/19 and Fig. IV/8), switch off this option to keep all synaptic events from the preceding simulation. If the synaptic events are not cleared the synapses in the new simulation database will behave as if the new simulation is a continuation in time of the preceding one. Improving simulation speed Many factors determine the duration of a simulation. With complex models, having hundreds of compartments or a lot of ionic currents, simulation speed can become an important limiting factor. An insight in what determines the simulation speed may prevent some problems. Some factors have a predictable, (semi)linear effect on simulation speed, they are described as correlations. The effect of several other factors is highly non-linear and difficult to predict because it depends on how fast other parameters in the simulation are changing ( stiffness, for example electrotonic length effects and the change in membrane potential). Hardware and System factors: - A fast computer yields of course a better simulation speed. The Macintosh computers on which Nodus 3 can run ordered by decreasing speed are: IIfx >> IIci > IIcx = IIx SE30 > II (1990 status). Upgrading from any Mac to a Mac IIfx will improve simulation speed a lot, the difference between the other Macs is in my opinion not significant (less than 30%). - A completely dedicated computer is faster than one that shares time between different applications. One can switch off the MultiFinder environment so that there is no time sharing on the Macintosh, but most users do not want to do this too often (and it is impossible in System 7). It helps however to Quit other applications if possible and to always run Nodus as the foreground application (so that its menu bar is shown). The use of virtual memory may also slow down Nodus because of frequent disk accesses. Simulation speed can be improved by an additional 10% (more if other applications are using the Mac) if the High multifinder priority option in the Preferences dialog is switched on. This however significantly slows down all other activities on the Macintosh and may produce some unwanted side-effects (be careful on networks). It may also increase the initial menu interface response time.
IV-48 Nodus Reference Model dependent factors: - Simulation speed is inversely correlated with the total number of compartments in the neuron model(s) and with the total number of subdefinitions. Ionic current subdefinitions or any Process dependent subdefinition are worse than others. See the Making New Models section for some tips on how to use different models for different experiments. - A very short electrotonic length of any compartment in the model may have catastrophic effects upon simulation speed (stiffness problem). - For ionic currents simulation speed is inversely correlated with the number of (in)activation factors. Fast ionic and synaptic conductance time constants can slow down the simulation. - A large number of active synapses (synaptic conductance larger than zero) in a network slows down simulation speed. Note that variable transmitter release synapses are almost continuously active. Integration dependent factors: - The accurate Fehlberg method is much slower than the hybrid Euler method. Use the Fehlberg method only for passive membrane models or to check final simulation results prior to publication. In both methods increasing accuracy (by decreasing the value typed into the Relative error or Maximum V box) can reduce the simulation speed a lot. - The major time step usually has little influence on simulation speed, but if it is too small it may limit the minor time step (because the minor time step can never be larger than the major time step) and slow down computation. Very large major time steps may slow down simulation speed when a lot of synapses are active in a network simulation. - The use of a table with precalculated values for conductance (in)activation factors increases computation speed in hybrid method integration of excitable membrane models much. The voltage range of the table should include all excitable membrane potential values computed during the simulation (otherwise the values have to be calculated repeatedly during the simulation). Do not make the voltage range larger than necessary; this increases the voltage step for the tabulated values with a loss of accuracy as consequence. - Once a synapse postsynaptic to a constant transmitter release site starts firing, it does not stop till the synaptic shut off time is reached (even when the conductance has become negligible small). Network simulations with a lot of active synapses can speed up considerable if a good synaptic shut off time is selected. However do not make the shut off time too small; errors may be introduced because large synaptic conductances will be shut off suddenly. - Swift changes caused by experiments (current steps, voltage clamps, firing of a synapse) can slow down the simulation when they occur, more so if the resulting change in membrane voltage has a large amplitude. While simulations are running the behavior of Nodus is completely determined by the time required to compute one minor time step (or several minor time steps if the.high multifinder priority option in the Preferences dialog is switched on). If this takes more than a second (for example with large neuron models, models with a lot of excitable membrane, Fehlberg method, etc.) then the interactive response to user commands (mouse clicks, selecting a menu command, etc.) will be sluggish. The same goes for other applications if Nodus is running in the background under MultiFinder. The only solution is to do nothing else on the computer during the simulation. Selecting Simulation Parameters An important design goal of Nodus 3 was to have user friendly access to all variables, this was realized with popup menus. The selection of simulation parameters requires the combined use of several of popup menus and the user has some control over how these popup menus are organized.
Nodus Reference IV-49 The selection of a simulation parameter Most simulation database parameters can be selected for graphic or text output (Fig. IV/22). The list of available simulation parameters includes values which are constant (like reversal potentials) or which are not available (like concentrations) in Nodus 3.1. These are included for future compatibility. In later versions of Nodus concentrations will be implemented and changing reversal potentials and maximum conductances will be possible. The first step in selecting simulation parameters for output is to choose the type of parameter. This is done with the value type selection popup menu (Fig. IV/22); this popup menu figures in the View/Edit Parameters, Configure Plots and Text Output commands. Simulation parameters which are not present in the simulation database are dimmed. Figure IV/22: the value type selection popup menu. With this menu a type of simulation parameter is selected for output. Usually a lot of parameters of a single type are available; the user has to select a specific parameter for output. In all the experiment commands the user also has to select parameters: for the Current Clamp and Voltage Clamp experiments compartments, for the Synaptic Firing Times command postsynaptic sites and for Block Ionic Currents experiments currents. All these output and experiment parameters are selected with the selection popup row (Fig. IV/23). A B C Compartment preselection Neuron selection Subdefinition selection Compartment selection Figure IV/23: three selection popup rows. In A a full set of 4 popup menus which determine a synapse parameter. In B a set of 3 popup menus which determine a membrane potential parameter. In C a popup row for an output parameter that is not in use. The parameter selection popup menus are created from the data present in the source model definition files. The selection popup rows function only if the source model definition files are present in memory and if they have not changed too much since the simulation database was created (see also the last subsection). The parameter selection process basically maps the model data structure to the simulation database structure.
IV-50 Nodus Reference Each selection popup row makes the unambiguous selection of a single simulation parameter possible. Active selection popup rows consist of a set of 3 or 4 popup menus depending on the selected value type (Fig. IV/23). Parameters are selected from left to right, i.e. first select the appropriate item in the first popup menu then in the second one, etc. The neuron popup menu will show either not used when nothing is selected or the local name of a neuron present in the simulation model. Select the appropriate neuron, the compartment popup menu will change to show the first (appropriate) compartment of that neuron. The compartment popup menu shows the compartment number and its name if it has one, otherwise its structure type. Select the desired compartment with the compartment popup menu (the subdefinition popup menu will change). If the value type is a membrane potential the selection is complete, otherwise a subdefinition selection has to be made also. The subdefinition popup menu shows a memory ID-number (which is of no concern to the user) for the subdefinition and its name. The use of the compartment preselection popup menu is explained in the next subsection. The compartment popup menus are created selectively, except for experiment value types. The compartment popup menu will only show compartments were the selected value type is present; for example if ionic currents are selected only excitable compartments will be shown. The neuron popup and compartment preselection popup menus are not selective, they show all available items. If the selected neuron or compartment preselection contain no appropriate compartments an alert warning about an empty selection will be shown (Fig. IV/31); the popup menus will revert to the previous setting. The selection popup rows for experiment value types are smaller. For injected currents they are similar to membrane potential selection rows with 3 popup menus (Fig. IV/23B). For voltage clamps only the neuron popup menu is shown. For conductance blocking factors the first popup menu is an ionic current popup menu and is the only popup menu shown. The use of selection popup rows differs slightly between output commands (Fig. IV/24) and experiment commands (Fig. IV/25). In output dialog windows several selection popup rows are present (Fig. IV/24); each row represents one parameter for graphic or text output. The selection popup row can be active (it shows 3 or 4 popup menus that determine a simulation parameter) or not (it shows not used). The number of parameters that will be plotted or saved to the disk is determined by the number of active popup rows. Active and inactive selection popup rows may be interspersed, though this does not improve clarity! Figure IV/24: the Configure Plots dialog window. Four selection popup rows are active, thus on the second axis four parameters will be plotted.
Nodus Reference IV-51 In experiment dialog windows there is only one selection popup row (Fig. IV/25), which is always active. It is used to select the place (i.e. a compartment, postsynaptic site or current depending on the type of experiment) on which the experiment settings in the dialog window will operate. All experiment settings can be active at several places at once; for example currents can be injected in several different compartments. In most experiment dialogs (except in the Voltage Clamp dialog) only one place in the model can be viewed; the Synaptic Firing Times dialog shows all firing times at one synaptic site, the Current Clamp command a single current injection in one compartment, etc. Use the selection popup row to select other places in the model and view or edit the experiment settings there. Initially the dialog window will show a place in the model where a setting is active or the first available place in the model if no experiment settings are active. All experiment commands have a button to Delete all the settings at all places in the model. Figure IV/25: the Synaptic Firing Times dialog window. Two firing times are preset at the postsynaptic site shown in the selection popup row; other firing times might be preset at other postsynaptic sites. The compartment preselection popup menu Popup menus are a nice instrument to select parameters, but they should not be too long. Imagine selecting a compartment for membrane potential output in a 1000-compartments neuron model, a compartment popup menu showing all 1000 compartments would be hard to use! Therefore Nodus never shows more than 50 items in a popup menu. To select which set of 50 compartments in a 1000-compartment model will be shown in the compartment popup menu, the compartment preselection popup menu is used. The compartment preselection popup menu is the second popup menu in the selection popup row (Fig. IV/27). It makes selection of a subrange in the compartment popup menu possible. The user has control over the compartment preselection process with 2 options in the Preferences command. The Select by structure type option in the Preferences dialog window forces Nodus to always display a structure type preselection popup menu (Fig. IV/26), corresponding to the Structure type popup menu in the compartment dialog window (Fig. IV/13) (with an all comps menu item added). The compartment popup menu will then only show compartments with the corresponding structure and where the requested value type is present. The all comps menu item will be dimmed for large neuron models. This option is useful if (most) compartments in the model have a structure type different from undefined. Note that in very large neuron models more than 50 compartments may have the same structure type. The Show only named comparts option in the Preferences dialog window affects the compartment popup menu directly. Only compartments which have a compartment name (Fig. IV/13) will be shown, unnamed compartments cannot be selected. This option is useful if compartments were named selectively, so that only interesting ones are shown. The Show only named comparts option may be combined with the Select by structure type option.
IV-52 Nodus Reference Figure IV/26: Compartment preselection by structure type. The complete structure type preselection popup menu is shown at the left. At the right the effect of a structure type preselection upon the compartment popup menu is shown in a membrane potential selection popup row. The appearance of the compartment preselection popup menu depends on the size of the selected neuron model and the Select by structure type option (Fig. IV/27). If the neuron model is small and no structure type selection is requested, the words all comps will be shown instead of a preselection popup menu (Fig. IV/24). If the neuron model is large and no structure type selection is requested, a range preselection popup menu is shown (Fig. IV/27). The range preselection popup menu just contains subranges of compartment numbers, i.e. #1-#50, #51- #52, etc. If the Select by structure type option is on, there will always be a structure type preselection popup menu. The size of the neuron model determines whether the all comps menu item is available or not. 5-compartment neuron: ionic current selection Select by structure type is OFF 109-compartment neuron: membrane potential selection no preselection popup menu Select by structure type is ON range preselection popup menu structure type preselection popup menu, all comps item is available structure type preselection popup menu, all comps item is dimmed Figure IV/27: four types of preselection popup menus are possible, depending on the size of the neuron model and the Select by structure type option. Error messages about parameter selection A simulation parameter selection maps the model data structure to the simulation database structure. This process will work fine if the model and database structures coincide, otherwise Nodus will warn the user that there is a problem. Which error message is shown depends on the circumstances. Errors during New Simulation A new simulation database inherits its output and experiment settings from the old simulation database (if present, otherwise defaults are used). If the model used for the new simulation is quite different from the preceding one, some output and/or experiment settings may no longer be relevant.
Nodus Reference IV-53 For example graphic output of ionic currents will not work very well for a passive membrane model. Nodus warns that some simulation parameters could not be found and lists them all in the error message (Fig. IV/28) Figure IV/28: error message during execution of a New Simulation command with a new source definition file. The new model is too different from the preceding one; some simulation parameters no longer exist. Correct the listed settings before running the simulation by executing the appropriate simulation output and/or experiment commands. When such an output or experiment command is activated an additional error message will be shown (Fig IV/30). If the settings are not corrected the simulation can run, but the involved output will be continuously zero and the concerned experiments will not work. Errors during simulation output or experiment commands The selection popup row in all simulation output and experiment dialogs is created from data in the source model definition files, which have to be in memory. If the source files are not in memory, Nodus will warn the user (Fig. IV/29). This condition can occur only after the user has Closed some definition windows or after an Open Simulation with the Open all linked files check box switched off (Fig. IV/8). Figure IV/29: error message warning that some source model definition files are not in memory. This message will displayed when a View/Edit Parameters, Configure Plots, Text Output, Current Clamp, Voltage Clamp, Synaptic Firing Times or Block Ionic Currents command is executed. The best solution is to press Cancel and Open the required source model definition files before executing the required command again. For output commands one can Continue. After Continue the output dialog windows will show the raw parameter selection numbers that Nodus uses internally. These selection numbers are not documented, this option is available for emergency use only. The average user should never edit the raw parameter selection numbers! Experiment commands are not accessible; the Continue button is replaced with a Delete experiment settings button.which clears the experiment completely when pressed If the source model definition files are in memory, but have been edited since the compilation of the simulation database, some of the parameter settings may have become invalid. The changes in definition files causing this problem are usually deletions: deleting a compartment (Fuse Compartments), removing a compartment tie to a subdefinition or removing a neuron in a network. Any output or experiment setting referring to the deleted compartment or subdefinition has then become irrelevant. Deletions can cause two problems: either simulation parameters are no longer available and an error message (Fig. IV/30) will be shown or some settings will seem to have shifted and point to different simulation parameters (see The initial value problem ).
IV-54 Nodus Reference Figure IV/30: error message displayed upon selecting an output or experiment command. Discrepancies between the simulation database and the (edited) source model definition files make the display of a parameter selection popup row impossible. Simulations will run without problems because the simulation database has not changed, but the parameter selection popup row can no longer display the correct popup menu item. Nodus warns the user which model structure was not found before displaying the simulation output or experiment command (Fig. IV/30). To prevent this problem one should keep original source model definition files, rename the file (and give it a new hidden ID-number) after large changes to the model structure by doing Save As. If Continue is pressed the raw parameter selection numbers that Nodus uses internally will be shown for those parameters that are no longer present in the source model files. Do not try to edit these parameter selection numbers, but change any other parameter or popup menu as you like. The Continue & delete concerned settings button is useful when this happened after a New Simulation database compilation. As the concerned parameters are no longer in the simulation database either one has to change or delete them anyway. All parameter selections that are no longer relevant will be replaced by the not used setting, the other selections remain unchanged. During popup menu parameter selections one may try to select a neuron or compartment preselection menu item that has no compartments with the requested value type. This could happen for example in a network simulation, when a synaptic current value type is selected and the user selects with the neuron popup menu (Fig. IV/23) a neuron that has only presynaptic sites. Then an alert like in Fig. IV/31 will be shown and the popup menus will revert to their previous setting. This may happen frequently if the Select by structure type option is active. Figure IV/31: error message warning that the selected neuron or compartment preselection popup menu item cannot be used for the requested value and structure type (in this case synaptic currents and neurites).
Nodus Reference IV-55
V-56 Nodus Menus V. NODUS MENU COMMANDS Apple Menu 5 5 File Menu 5 5 Edit Menu 6 2 Simulation Menu 6 4 Windows and cursors...64 Simulation value types...66 Menu commands...67 Network Menu 8 0 Windows...80 Menu Commands... 81 Neuron Menu 8 3 Windows...83 Menu Commands... 87 Conductance Menu 9 9 Windows...99 Menu Commands...100 Apple Menu About Nodus The About Nodus dialog window shows the Nodus version and the available free heap memory in kilobytes (K) and % of the total heap memory. Note that the global memory containing all the definition data and the simulation database is not included in this count (see chapter II). Available memory should be higher than 100K, if it gets below 80K Nodus will alert the user whenever a menu command is invoked. Figure V/1: the About Nodus dialog window. File Menu The Nodus menu is similar to the File menu in most Macintosh applications. It contains all the menu commands used to manage Nodus files. Some menu items contain submenus, see chapter IV and Fig. IV/2; submenus are listed here as separate menu commands.
Nodus Menus V-57 Figure V/2: the File menu. New Simulation This menu command is enabled if there is no simulation database in memory and if an old database is still available and/or a network definition file or neuron definition file is in memory. Figure V/3: the New Simulation command dialog window. The New Simulation command dialog window contains 2 section: one specifying the source model definition file for the New simulation database and one setting the Initial values. Refer to chapter IV, Fig. IV/19 for a more extensive discussion on simulation database compilation and the initial value problem. Press OK to create a new simulation database. Simulation database: The new simulation database can be either: Use old simulation database : an exact copy of the simulation database used in the preceding simulation. Compile from network definition file : a new simulation database is compiled from the network definition, with all the parameters taken from the network definition window or from memory if the window is hidden. Compile from neuron definition file : a new simulation database is compiled from the neuron definition, which can be selected with a popup menu. All parameters are taken from the neuron definition window or from memory if the window is hidden. Initial values: Values in memory: all integrated simulation parameters are taken from the old simulation database. Values which cannot be found are initialized to equilibrium values for resting potential. Resting potential: all compartmental membrane voltages are set to their neurons resting membrane potential. Other integrated values are put at equilibrium values for resting potential.
V-58 Nodus Menus Set to mv: all compartmental membrane voltages are set to the user specified membrane potential. Other integrated values are put at equilibrium values for this potential. Which option is preselected depends on whether an old simulation database exists and on the settings in the Preferences command. Additional initial values options are: Clear all synaptic events: any synaptic events still in memory from preceding simulations will be cleared. Clear experiments & text output: all experiments and text output settings in memory are disabled. No experiments or text output will be active. New Simulation This menu command is available when the shift key is pressed. No dialog window will be shown if there is an old simulation database in memory; the Use old simulation database and Values in memory options are used. New Network This menu command is enabled if there is no network definition file in memory and if at least one neuron definition file is in memory. The New Network command creates an empty network definition window (Fig. IV/14) called Untitled Network. See the Network Menu section for a description of the window contents. New Neuron This menu command is always enabled (unless 20 neuron definition files are in memory!). The New Neuron command creates a default neuron definition window (Fig. IV/12) called Untitled Neuron #. See the Neuron Menu section for a description of the window contents. New Conductance This menu command is always enabled (unless 20 conductance definition files are in memory!). The New Conductance command creates an empty conductance definition window (Fig. IV/11) called Untitled Conductance #. See the Conductance Menu section for a description of the window contents. Open Simulation This menu command is enabled if there is no simulation database in memory. Figure V/4: the Open Simulation file dialog window. A standard minifinder dialog window is shown that lets the user select a simulation data file to Open. A Simulation Plot Window will be displayed. Options: Clear all synaptic events: any synaptic events present in the simulation data file will be cleared.
Nodus Menus V-59 Open all linked files: the source model definition file and all files linked to it are opened also (unless they are already in memory). See the Nodus Files section in chapter IV for more details. This option should always be used; the simulation output and experiment commands will not work properly if the source model definition files are not in memory. Open Network This menu command is enabled if there is no network definition file in memory. Figure V/5: the Open Network file dialog window. A standard minifinder dialog window is shown that lets the user select a network definition file to Open. A network definition window (Fig. V/30) will be displayed. See the Network menu section for a description of the window contents. Options: Open all linked files: all the neuron definition files necessary for the network and all files linked to them are opened also. See the Nodus Files section in chapter IV for more details. This option should always be used. A network definition cannot be loaded into memory if one of its neuron definition files is not in memory. Open Neuron This menu command is always enabled (unless 20 neuron definition files are in memory!). Figure V/6: the Open Neuron file dialog window. A standard minifinder dialog window is shown that lets the user select a neuron definition file to Open. A neuron definition window (Fig. V/33) will be displayed. See the Neuron menu section for a description of the window contents. Options: Open all linked files: all the conductance definition files necessary for the neuron are opened also. See the Nodus Files section in chapter IV for more details. This option should always be used. Open Conductance This menu command is always enabled (unless 20 conductance definition files are in memory!).
V-60 Nodus Menus Figure V/7: the Open Conductance file dialog window. A standard minifinder dialog window is shown that lets the user select a conductance definition file to Open. A conductance definition window (Fig. V/52) will be displayed. See the Conductance menu section for a description of the window contents. There are no options. Import Simulation This command is not implemented in Nodus 3.1. Import Neuron This menu command is always enabled (unless 20 neuron definition files are in memory!). Figure V/8: the Import Neuron file dialog window. At right the File format popup menu. A standard minifinder dialog window is shown that lets the user selects a standard Macintosh text file containing morphological data to Open. The contents of this morphology file are interpreted by Nodus and converted into a neuron definition. A default neuron definition window (Fig. V/12) will be displayed. See the Neuron menu section for a description of the window contents. Options: File format: this popup menu specifies the format of the morphological data. These formats are defined in the Appendix. The Import Neuron command will not work if a wrong file format was selected. Keep 3-dimensional coordinates: the 3-dim coordinates present in the file are stored into the neuron definition file. This option has no advantages in Nodus 3.1, it reduces the maximum number of compartments from 4000 to 3000. Automatic compartment names: compartment names are generated automatically. These names reflect morphological information, like branching order. Do not use this option if the Select only named comparts option in the Preferences dialog is set, because all compartments will be named.
Nodus Menus V-61 Close a window Is enabled if any window is displayed. If it is a definition window, its contents are checked and stored in memory if no errors were found. If any parameter value in the window is not acceptable, Nodus will alert the user and the window will not be closed. If the contents of the window were changed and these changes were not Saved, Nodus will alert the user before closing the window (if the appropriate Warnings option is active in the Preferences dialog). Figure V/9: Warning shown before a window is Closed that has been edited, but not Saved. Press Yes to save the changes and close the window; No to close it without saving; if Cancel is pressed the window is not closed and not saved. The action of the Close command depends on whether the window belongs to a linked file or not. If it is a file linked to other files in memory (see the Nodus Files section in chapter IV for more details), the window will be hidden. Otherwise the window will be closed and all its data will be removed from memory. Hide a window This command is available when the shift key is pressed and if any window is displayed. If it is a definition window, its contents are checked and stored in memory if no errors were found. If any parameter value in the window is not acceptable, Nodus will alert the user and the window will not be hidden. The window data remain in memory after the window is hidden. The window can be displayed again by selecting the menu item with the same title at the bottom of the appropriate menu. Kill a window The command is available when the shift and the option key are pressed while a window is present. Kills the window, i.e. the window is closed without checking it contents. Any changes made to the window contents are not stored. This command is useful to quickly close a window that has several bad parameter values in it. Close All Is enabled if any window is displayed. Closes all Nodus windows in the same way as consecutive Close a window commands. Refer to Close a window for more details. Hide All This command is available when the shift key is pressed and if any window is displayed. Hides all Nodus windows in the same way as consecutive Hide a window commands. Refer to Hide a window for more details. Close All Graphs This command is available when the option key is pressed and if any window is displayed. Closes all graphic windows, i.e. old simulation plots (not the active Plot Window), Neuron Diagrams and Conductance Plots. There is no option to save these windows first.
V-62 Nodus Menus Save a window Is enabled if any definition or simulation window is displayed. The window contents are checked and stored in memory if no errors were found. If any parameter value in the window is not acceptable, Nodus will alert the user and nothing will be saved. The contents of the corresponding definition file are updated. The name and hidden ID-number are not changed (see Fig. IV/6). If the file has never been Saved before (still called Untitled ) the Save As dialog (Fig. V/10) is presented. Save As Is enabled if any definition or simulation window is displayed. Figure V/10: the Save As file dialog window for conductance (left) and neuron (center) definition files. At right the File format popup menu. Presents the standard minifinder dialog window which lets the user supply a new name and/or folder location for the file. Press Save to confirm the command. A new file will be created; it will have the new name and a new hidden ID-number. The old file is not changed. Options: Update all links in memory: all links in hierarchically higher files in memory that point to this file are updated to link to the new file instead of the old file. See the Nodus Files section in chapter IV and Fig. IV/6 for more details. File format (neuron definition files only): neurons can either be saved in the Standard Nodus format (default, this is the only format that can be Opened) or some of the data can be written to a text file. A text file can be either an Import Neuron File Format (Ladder or Genesis, see the Appendix ) or it can be similar to printed output (Text). Page Setup Presents the standard Macintosh page setup dialog, specific for the active printer. Use this command to specify the page size and any reduction or enlargement. The settings are kept in memory till Nodus quits. The 50% Reduction option is very useful if the Double size simulation plots option in the Preferences dialog is selected. Print Is enabled if any definition or graphics window is displayed. The standard Macintosh print dialog, specific for the active printer, is presented so that the user can select the number of copies, the quality of printing, etc. The contents of the window are printed, the date and the window title are printed in a header. If it is a graphics window the graphic is printed as it is displayed. Do not use draft printing for graphics! If it is a definition window the complete definition is printed, using the Print font specified in the Preferences dialog. Note that printing of large neuron definition files may take a lot of time. No dialog is shown during the preparation or the actual printing. Actual printing can be interrupted by pressing command-period.
Nodus Menus V-63 Automatic Saving The Automatic Saving feature makes Nodus save the simulation database with all integration, output and experiment settings and the simulation results (including the Plot Window) in the Nodus Resume file on Quitting. When Nodus is restarted, the application will look for this file and load all data so that Nodus behaves as if it had never been interrupted. An instruction window is displayed if a simulation database present: Figure V/11: the Automatic Saving instruction window. This command is a toggle: a check mark is shown when the automatic loading/saving feature is on, select the command again to turn it off. The automatic command saves only the simulation database, definition files are not saved automatically (but the appropriate warnings will be displayed before Quitting). Quit When this command is selected Nodus closes all windows and output files, writes a Nodus Resume file if the Automatic Saving command is on, and returns to the Finder. Nodus will check whether all definitions and the simulation database have been saved correctly if the appropriate Warnings option is active in the Preferences dialog. An alert will warn the user if any data have not been saved correctly, see Fig. V/8 (the Cancel button interrupts the Quit command). Edit Menu The Edit menu has a limited function in this Nodus version. It is used to transfer graphics to other applications and it is used by desk accessories. Figure V/12: the Edit menu. Copy The only Nodus specific edit command is enabled when a graphics window (i.e. the Simulation Plot Window, an old simulation plot, a Neuron Diagram or a Conductance Plot) is the active window. It copies the graphic to the Clipboard in PICT format. The graphic can then be pasted in the Scrapbook or in other application windows (like MacWrite or MacDraw ). Undo, Cut, Paste, Clear These commands work only with some desk accessories. Refer to Macintosh, the owners guide for further information.
V-64 Nodus Menus Preferences This command presents a dialog window to let the user change several default settings in Nodus. These settings are stored in the Nodus Preferences file. Figure V/13: the Preferences dialog window. General options: Tile windows: all definitions (except alerts) are tiled from the top, left of the screen to the bottom, right. Otherwise the windows are centered on the screen. Default is: on. High multifinder priority: Nodus gives the running simulation top priority for computer time while Nodus is running in the foreground. Simulations can compute up to 3 seconds before any other activity is allowed. This means that other applications running in the background will almost stop, which may be unacceptable (for example network activities or print spooling). The response of Nodus to mouse clicks may also be sluggish if no menu commands have been executed for some time. Default is: off. Print font: selects the font used for printing of definitions and for text in graphics windows. Printing works best if Geneva or Helvetica is selected. If a Postscript laser printer is used, an appropriate font like Helvetica should be selected. Default is: Geneva. Simulation options: Automatic loading/saving: this is identical to the Automatic Saving option in the File menu. Default is: off. Beep when finished: the computer beeps when the simulation has finished running. Default is: on. Quit when finished: Nodus Quits when the simulation has finished running. This may be friendly to other users in a multi-user environment. Default is: off. Double size simulation plots: new simulation plots are drawn in double resolution. They will not fit on most Macintosh screens, but they will look nicer if printed at 50% Reduction (Page Setup dialog in the File menu). Default is: off. Current plots: negative up: all ionic and synaptic currents will be plotted with outward currents (negative values) going up and inward currents (positive values) going down. This is the voltage clamp display that may be more familiar to electrophysiologists than the standard display used by Nodus. Default is: off. Default to values in memory, Default to resting potential, Default to set potential: selects which initial values option will be the default in the New Simulation dialog window (Fig. IV/19). Default is: Default to resting potential. Reset simulation numbers: resets the simulation number counter to zero. Cannot be undone.
Nodus Menus V-65 Warnings options: Before closing definition file: Nodus will display a warning when any definition window that has been changed without saving is going to be Closed. Default is: on. Before closing simulation file: Nodus will display a warning when a Simulation Plot Window of a simulation database that has been changed without saving is going to be Closed. Default is: off. Neuron definition options: Multiple use of subdefinitions: subdefinitions may be used by more than one neuron definition file. See the Nodus Files section in chapter IV for more details. Default is: off. Only node connections: the branch connection and weight factors options for links between compartments are not displayed in compartment definition dialogs. Not available in Nodus 3.1. Default is: off. Select by structure type: the compartment preselection popup menu in selection popup rows will always be a compartment structure type preselection popup menu. See Selecting Simulation Parameters in chapter IV for more details. Default is: off. Show only named comparts: the compartment popup menu in selection popup rows shows only compartments that have a compartment name. See Selecting Simulation Parameters in chapter IV for more details. Default is: off. Conductance options: in ns, in ms/cm 2 : all Gmax values in new Ionic Currents or Synaptic Currents subdefinitions will be specified in the selected unit. The setting in existing subdefinitions cannot be changed. Default is: in ms/cm 2. Simulation Menu Windows and cursors Plot Window The simulation results are always plotted in a Plot Window. The plot settings are under complete user control with the Configure Plots command, which can only be used before the simulation has Run. Plots are not drawn continuously and may look staggered ; this is because Nodus tries to draw complete lines instead of individual points (see the Update Plots command for more details). Figure V/12: the Plot Window.
V-66 Nodus Menus The Plot Window has a grow box and scroll bars. The grow box is always active, but the scroll bars remain dimmed till the simulation has finished running. Like all simulation windows, the Plot Window enables the Simulation menus. Closing the Plot Window is equivalent to closing the simulation database. The simulation database can be closed however without closing the Plot Window: press the option key while clicking in the close box or executing the Close simulation command. The Plot Window becomes a graphic window. It remains listed at the bottom of the Simulation menu, but will no longer enable Simulation menu commands. Time Window The Time Window shows the simulation time in milliseconds. This time will change continuously while the simulation is running. Figure V/13: the Time Window. Like all simulation windows, the Time Window enables the Simulation menus. The Time Window is optional, it may be closed without affecting the simulation database status. Just select the corresponding command in the Simulation menu to get a new Time Window. Measure Window The Measure Window shows the position of the crosshair cursor in the Plot Window, transformed in time and value measurements in the correct units (Figs. V/14 and V/29). Figure V/14: the Measure Window. Each plot in the Plot Window has its own (rectangular) measuring region. If more than one plot is drawn then the values and units shown in the Measure Window will depend on the scaling of the axis in the measuring region, were the cursor is located. If the cursor is outside the Plot Window, the display in the Measure Window does not change anymore. If the mouse button remains pressed and the cursor is dragged over the Plot Window the Measure Window will show the differences between begin and end point of the cursor movement and the slope of the curve. Drag the cursor to measure period of firing, size of an action potential, rising slope of an IPSP, etc. Like all simulation windows, the Measure Window enables the Simulation menus. Click the close box of the Measure Window to stop measuring, this will not affect the simulation database status. Status Window The Status Window shows information about the actual simulation computations. It can be used to time total computation time or to try to locate trouble spots in a simulation model that slow down computation. Started on: date and time on which the simulation was started. Integration method: hybrid or Fehlberg method. Computation time: total computation time in minutes till now. Only actual computation time is counted, Pausing or interrupting the simulation will not affect the computation time (except the time necessary to generate a new equation table in the hybrid method). Computation speed: the average computation speed in simulated milliseconds versus real time minutes.
Nodus Menus V-67 Figure V/15: the Status Window. Relative error/maximum V: actual maximum relative error (Fehlberg method) or maximum change in membrane potential (hybrid method) used by the integration routines. Estimated error: the estimated error will only be shown if the simulation is Running. What is shown depends on the integration method chosen. For the hybrid method it will be the largest change in membrane potential during the last integration step, for the Fehlberg method it will be the estimated relative error. If the estimated error is shown bold the integration step has failed, the minor time step will be decreased and a new attempt at integration will start. If it is shown plain the integration step succeeded. The simulation database array index at which the largest error occurred is shown. Minor time step: the actual minor time step as determined by the integration routines. Like all simulation windows, the Status Window enables the Simulation menus. Its slows down simulations a lot, so use it only when needed. Closing the Status Window does not affect the simulation database status. Run cursor While a simulation is running, the cursor changes to the run cursor: a small oscilloscope. Two control lights indicate which integration method is being used (see Integration Settings). The left, red light burns when the hybrid method is used; the right, blue light burns when the Fehlberg method is used. Figure IV/16: the run cursor. Left: hybrid method used, right: Fehlberg method used. Use the center of the oscilloscope screen to point with the run cursor. The appearance of the run cursor does not depend on what type of window is active. Crosshair cursor The crosshair cursor is shown when measuring is active (Measure Window). Figure IV/17: the crosshair cursor. The center of the crosshair is the point being measured. Simulation value types To select parameters for output (in the View/Edit Parameters, Configure Plots and Text Output commands) one has to specify the type of value with a popup menu (Fig. V/18, Selecting Simulation Parameters in chapter IV). Some of these value types are not the original model parameters, they have been optimized for simulation speed. The exact definition of each value type follows, refer to chapter III for more information about the equations defining these variables.
V-68 Nodus Menus Figure V/18: the value type selection popup menu. - membrane potential: E (Eq. 4, Eq. 28). - conductance (in)activation: M or H (Eq. 21). - conductance time constant: τ M or τ H (Eq. 23). - ionic conductance: G(E,t) (Eq. 24-26). - maximum conductance: derived from G max (Eq. 24-26). Has been scaled for membrane surface (if it was in ms/cm 2 ). - reversal potential: E k (Eq. 27). - ionic current: I k (Eq. 27). - transmitter release: if variable: T s (Eq. 29-31); if constant a derived value: -b (synapse is not firing) or b (synapse is firing). - synaptic conductance: G s (Eq. 32-36). - synaptic peak conductance: derived from G max (Eq. 32-36). Has been scaled for membrane surface (if it was in ms/cm 2 ). For dual exponential (Eq. 34) functions it contains the normalizing factor for the variable conductance. - synaptic reversal potential: E s (Eq. 37). - synaptic current: I s (Eq. 37). - concentration: the concentration in its local units (not available) - in/out process: the diffusion, buffering, etc process in its local units (not available). - injected current: I e (Eq. 4, Eq 28). - voltage clamp current: I v = -CM n E/dt. - conductance blocking factor: varies between 0 and 1; 1 is no block, 0 is complete block. Menu commands The File menu commands are active when a simulation window is the highlighted window. Several File menu commands can be italicized (Chapter IV) when the simulation has started or finished running. This means that the command will show a dialog window that can be viewed, but not edited. The user can then examine the integration, output and experiment settings but not change them (the simulation has finished!).
Nodus Menus V-69 Figure IV/19: the Simulation menu Run The command is enabled when a simulation database is in memory that has not yet finished running. Starts or continues the simulation and the command name changes to Pause. The cursor changes to the run cursor (a small oscilloscope) while the simulation is running. If the hybrid Euler method is selected (Integration Settings), a equation table will be created first and a dialog window will show progress information. See Making New Simulations in chapter IV and Fig. IV/17) for more details on how the Run command affects the Simulation menu. Running the simulation is an iterative process: the integration routine calculates the membrane potential and concentrations in all compartments and all other relevant parameters and tests if the results meet the accuracy requirements. If the test fails the minor time step is decreased and the integration routine tries again (this process can be followed in the Status Window). If the integration step is successful the simulation time is increased by the minor time step (visible in the Time Window) and plot data are send to the Plot Window; then another integration step is started unless the simulation is finished. Pause While a simulation is running the Run command changes into the Pause command. Pauses the simulation and the command becomes Run again. Pausing the simulation has limited use. It is the only way to access the Integration Settings command after the simulation has started. One can also Pause a simulation if Nodus is interfering too much with other applications which are running under the MultFinder, continue Running when the other applications have finished. Integration Settings The command is enabled if the simulation has never Run or during Pauses. It is italicized when the simulation has finished running. This command gives control over the integration procedures used by Nodus to calculate the changes in membrane potential and concentrations in the compartments. The settings can be changed any time before or after a simulation has started, usually the integration method will not be changed during a simulation. Time controls Set the duration and the minimum number of time steps for the simulation: Begin: time in milliseconds at which the simulation will start, or the current time after a simulation has started. The default setting is 0 ms, most users will never change this value.
V-70 Nodus Menus End: time in milliseconds at which the simulation will stop. This important value sets the simulation length, it depends on the experiment that is being simulated (default: 1000 ms). Major time step: the integration routines are forced to adapt the minor time steps so that each major time step is reached exactly. Experiments can start or stop only at major time steps and Text Output may also be controlled by the major time step. See Integration Methods in chapter III and Making New Simulations in chapter IV for more details on minor and major time steps. The default setting is 1 ms. Fixed time step: integration will be performed with a fixed time step, equal to the major time step. The integration methods normally use variable time steps, which is more efficient. In exceptional situations the step adaptation routines may fail and a fixed time step has to be used. Note that Relative error/maximum V checking will have no effect, the simulation will fail when membrane potential exceeds ±200 mv. Synaptic currents Shut off after: the synaptic shut off time. Synaptic conductances controlled by alpha or dual exponential functions will be switched off after the specified time. See Making New Simulations in chapter IV for more details. The default setting is 1000 ms. Figure V/20: the Integration Settings dialog window. Left: hybrid Euler method is selected, right: Fehlberg method is selected. Integration method Controls which integration method will be used. Hybrid Euler method: a fast but relatively inaccurate integration method. Maximum V: this value controls the accuracy of the hybrid method. It sets the maximum amount by which the membrane potential may change in any compartment during one minor time step. The integration routines will decrease the minor time step till this criterion is fulfilled or give an error message if the minor time steps gets too small. Default is 0.1 mv. Maximum [ ]: additional control of the accuracy of the hybrid method. Is shown when concentrations are present in the model. Sets the maximum amount by which any concentration may change during one minor time step. Equation table from to mv: to increase computation speed a table of conductance (in)activation factors is calculated before the simulation starts. The range that is precalculated is determined by this setting, outside this range the (in)activation factors have to be computed during the simulation. See Integration Methods in chapter III for more details. Default is from -80 to 45 mv Fehlberg method is a fifth order Runge-Kutta method, it can be very accurate (up to 10-12 ), but is sometimes slow. It is quite sensitive to the stiffness of the differential equations.
Nodus Menus V-71 Relative error: this value controls the accuracy of the Fehlberg method. It sets the maximum estimated relative error during the integration of membrane potentials, concentrations, variable transmitter release, synaptic conductances and ionic conductance (in)activation factors in any compartment during a minor time step. The integration routines will decrease the minor time step till this criterion is fulfilled. View/Edit Parameters The command is always enabled if a simulation database exists. It is italicized when the simulation has finished running. Figure V/21: the View/Edit Parameters dialog window. With this command one can view most of the simulation database parameters. One can also edit the simulation parameters to change their (initial) values, but be very careful with this option. Editing membrane potentials is easy, editing anything else may produce unexpected results! Parameters selection popup row: The box in the upper part of the dialog window contains a value type selection popup menu (Fig. V/18) and a parameter selection popup row (see Selecting Simulation Parameters chapter IV). These allow selection of one simulation parameter to view or edit. Value: Value: the value of the selected simulation parameter. V/ t: the differential of the selected simulation parameter. Is not available for all parameters. Allow editing: check this option to be able to edit the selected parameter, but beware! Any changes are stored in the simulation database when OK is pressed or another simulation parameter is selected with the popup row. Configure Plots The command is always enabled if a simulation database exists. It is italicized once the simulation has started running. This command gives a choice between 7 kinds of plotting configurations, with up to 4 separate axes and a maximum of 20 plotted parameters. Configure the plot settings before running the simulation, afterwards the settings cannot be changed anymore. Axis selection The left upper corner contains 7 plot configuration icons. The configurations differ in the number and position of the plot axes available. Click on one of the icons to select it, it will be highlighted. Edit axis #: these push buttons controls which axis data are shown in the lower part of the dialog. Press a button to change the displayed axis data, any changes made to the presently shown axis data will be stored in memory.
V-72 Nodus Menus Figure V/22: the Configure Plots dialog window. Plot options The upper right part of the dialog contains some options which affect all axes. Draw axes: if this option is checked time and value axes are drawn for all the plots (default). The time and value axes cross if zero is included in the value range, else a hanging time axis is drawn. If this option is not checked a scale marker is placed at the lower right corner for every plot. Repetitive sweeps: the graphic output is repeated in the same plotting areas in sweeps. When the end of the first Time axis is reached, all plots continue at the start of their Time Axis. All plots are drawn in the same color (on color screens), the color changes for each new sweep. This option is particularly useful for Voltage Clamp experiments. 1 time axis range: all displayed axes will have the same Time Axis range. Changing the Time Axis range for one axis, automatically changes it for all other axes also. This is what one usually wants; all plots are aligned in time. Axis data: The lower part of the dialog contains the data about one axis, its number is shown at the top. Press the Edit axis # buttons above to change the displayed axis data. Time axis: enter the begin and end of the time axis in the from to text boxes (default is the total simulation time). The time unit can be changed with the unit popup menu. The time limits set in Integration Settings can be exceeded, this can be used to have the same scale for graphic output of simulations with different total simulation times. The time scale can be changed to enlarge significant events, etc. Value axis: select the displayed value type with the value type popup menu (Fig. V/18). Each axis can display only one value type. Enter the begin and end of the value axis in the from to text boxes. The value unit (which is specific for the selected value type) can be changed with the unit popup menu. Choose values which are slightly larger than the maxima and minima reached by the plotted compartments so that the plots are 'filled. Significant events can be enlarged by choosing smaller value ranges, but a part of the plot will be lost. Below the axis data are 5 parameter selection popup rows (see Selecting Simulation Parameters in chapter IV) which are used to specify which parameters will be plotted. Each row corresponds to one plotted value, the row number is shown in the correct plot color (on color screens). Update Plots The command is enabled while a simulation database is running.
Nodus Menus V-73 Figure V/23: part of the Plot Window before (left) and after (right) a Update Plots command. This commands updates all plots so that the most recent computation results are shown. While a simulation is running the graphics in the Plot Window sometimes do not reflect the actual integration status. Nodus tries to decrease the memory use of plots by converting the stream of points generated by the integration routines into continuous lines whenever possible. This is accomplished by buffering a small amount of points before plotting them. When several plots are drawn they may seem to stop at different time values. This does not matter if no one is watching the screen during the simulation. To check the results intermittently during a simulation select Update Plot to flush the plot buffer so that all plots show exactly how far the simulation has progressed. Text Output The command is always enabled if a simulation database exists. It is italicized if Text Output is active, once the simulation has started running. A check mark shows whether it is active. Figure V/24: the Text Output dialog window. This command controls the output of simulation results to a text output file on disk. The output settings can be selected either before a simulation has started or while it is running, but once Nodus has started to output the settings cannot be edited anymore. If text output is active then at the next Run command a minifinder window prompts for the text filename (default is the simulation name + results ). The text file is always saved to the same folder (or the desktop level) as the Nodus applications, the user can specify another folder but it will have no effect.
V-74 Nodus Menus Delete all: clears all Text output settings in the simulation database. Text output is no longer active. Output options: One of two output modes can be selected in the upper left part of the dialog. Output at major time step: at the end of each major time step (see Integration Settings) the results are written to the file. Output consists of the time in ms, followed by the selected simulation parameters. Warning: this option can create huge disk files. Output maxima/minima: can be used to time action potentials, EPSPs, IPSPs, etc. Nodus checks at the end of each minor integration step if the selected simulation parameter has reached a maximum or minimum and outputs the parameter array index, the time of occurrence (in ms) and the value of the parameter if this is the case. Either maxima or minima alone or both together can be monitored. Because of the additional calculations involved, this command decreases the integration speed slightly. Threshold: this text box appears only for the Output maxima/minima option. Integration methods may oscillate slightly (see Integration Methods in chapter III), producing many fake maxima and minima. To restrict output to the real maxima and minima a threshold value has to be specified; this specifies how much the parameter value has to change in % versus the tentative maximum/minimum before it is considered a real maximum/minimum. Usually large threshold values (like 50 to 100%) will give satisfactory results, but one may have to experiment. Trigger all output on value #1: this check button appears only for the Output maxima/minima option. If it is checked only the first selected simulation parameter will be monitored for maxima/minima. All following simulation parameters will be send to output together with the first one whenever the first one reaches a maximum or minimum. Format options The upper right part of the dialog contains some options which affects all output. Separated by TABs: if this option is checked all simulation parameters will be written as floating point output separated by tabs, otherwise they are separated by spaces. This option is useful if one wants to import the text file into a spreadsheet, graphics application or a statistics package. Make legend: a legend describing for each array index the selected simulation parameters is put at the start of the text file. Not available in Nodus 3.1.2. Output data: The lower part of the dialog contains the simulation parameters selected for output. The dialog can show selected simulation parameters for only one value type at once, but several value types can be send to output. The different output value types are stored consecutively and one can go back and forth between them. Previous value type: one can use this button to look at other value type selections when more than one value type has been selected for output. The button is dimmed if one is looking at the first output selection (row numbers start with one). Pressing this button stores all changes made to output of the presently shown value type in memory. Next value type: use this button to select or view additional output of another value type. One can also use this button to enter more than 5 simulation parameters of the same value type. It will be dimmed if no simulation parameters are selected in the dialog for the present value type. Pressing this button stores all changes made to output of the presently shown value type in memory.
Nodus Menus V-75 Value type: Select the output value type with the value type popup menu (Fig. V/18). If some simulation parameters have already been selected for this value type, they will be shown. Changing the value type stores all changes made to output of the presently shown value type in memory. The value Unit (which is specific for the selected value type) can be changed with the unit popup menu. Output all values of this type for selected neuron: instead of one (selected) simulation parameter, all simulation parameters of the same value type present in the same neuron model will be send to output. Only available for Output at major time step. Beware of the huge disk files! Below the value type data are 5 parameter selection popup rows (see Selecting Simulation Parameters in chapter IV) which are used to specify which simulation parameters will be written to output. Each row corresponds to one simulation parameter selection. The row number is the number of the selected parameter in the list of all output selections (it can be any number between 1 and 20). Row numbers are empty till an output selection is confirmed. Current Clamp The command is always enabled if there is a simulation database in memory. It is italicized when the simulation has finished running. A check mark shows whether it is active. Figure V/25: the Current Clamp dialog window. Right: alternate views for white noise currents (upper right) or non-cyclical currents (bottom right). With this command current injections in one or several compartments of the neuron model can be specified or changed. Multiple current injections in the same compartment are possible, if they occur simultaneously they will be added together. Use this feature to make complex current combinations. The Current Clamp settings can be edited while a simulation is running. This causes no problems, retroactive changes are of course not possible. Delete all currents: clears all Current Clamp settings in the simulation database. Current Clamp is no longer active. Compartment selection: The box in the upper part of the dialog window contains a parameter selection popup row (see Selecting Simulation Parameters chapter IV) of the membrane potential type (3 popup menus). These allow selection of a compartment to inject currents into, any changes made to the presently shown current injection will be stored into memory. If no Current Clamps are defined the first compartment in the simulation database will be shown, otherwise the first compartment with current injections.
V-76 Nodus Menus Delete currents in this compartment: all current injections for the compartment shown are cleared. Current injections in other compartments are not affected. Show all compartments: the compartment selection popup menu shows all compartments in the selected neuron (default). If this option is not checked, the popup menu will only contain compartments for which current injections have been defined; this is a quick way to identify such compartments in large models. The lower part of the compartment box has a graphic display of all the current injections defined in this compartment. The time axis of the current display is the total simulation time as defined in the Integration Settings, the vertical axis is scaled to show the maximum current. Individual current injections are shown light blue, the sum of all currents in the selected compartment dark blue, and zero current is shown as a broken red line. The current display is not updated continuously, click on the display to update it. The interpolation accuracy of the display (and the time needed to update it) depend on the major time step (see Integration Settings). Current selection: Several different currents can be injected into a compartment simultaneously or consecutively. Currents that are injected simultaneously will be added together. Only one current injection can be viewed and edited at once. Different current definitions in the injection list in one compartment are numbered consecutively. Previous: shows the preceding current definition in the injection list for this compartment, dimmed if current # 1 shown. Pressing this button stores all changes made to the presently shown current injection in memory. Next: shows the next current definition in the injection list for this compartment, or an empty current definition if no next injection is defined. Pressing this button stores all changes made to the presently shown current injection in memory. Duplicate: duplicates the presently shown current definition, puts it at the end of the injection list and displays it. Dimmed if the current definition is empty. Delete: deletes the presently shown current definition and shows the next one in the injection list for this compartment. Dimmed if the current definition is empty. The rest of the buttons and text boxes define the presently shown current definition. The icons at the upper left determine the type of current. They are (from left to right): steady current, ramp current, triangular current, sinusoidal current and white noise current. A single unit of current definition corresponds exactly to what is shown in the icon. Begin: time in ms at which the current injection will start. This time has to be smaller than the maximum simulation time (see Integration Settings) and has to be a multiple of the major time step (idem). End: time in ms at which the current injection will end. This time has to be smaller than the maximum simulation time (see Integration Settings) and has to be a multiple of the major time step (idem). Amplitude: (maximum) amplitude of the current in nano-amperes (na). If no value is specified, the current definition is considered empty. Corresponds to the amplitude of steady currents and the maximum for range and triangular currents. Sinusoidal and white noise currents fluctuate between plus and minus the amplitude. Cyclical: if this option is not selected the defined current unit will be active during the simulation from the Begin till the End time. If cyclical is selected, the current definition unit will be repeated several times in between the Begin and End time, the frequency of repetition depends on the specified Period. Default is off.
Nodus Menus V-77 Period: the period in ms for a Cyclical current definition. The current definition unit will have a duration of one period and will then be repeated till the End time is reached. This value is not shown when Cyclical is not selected. The period has to be smaller than the maximal simulation time (see Integration Settings) and it has to be a multiple of the major time step (idem). Default is from Begin to End (one period). On from to : the part of the Period in ms during which the current injection is active. The main use of this feature is to create repetitive current pulses with a cyclical steady current; the Period determines its frequency, the on from to values determine the length of the pulse. They can also be used with non-steady currents to produce even more complex current injections. These values are not shown when Cyclical is not selected. The on from to values have to be smaller than the Period and they have to be a multiple of the major time step (see Integration Settings). Default is the full Period. Seed: with white noise currents a seed number for the random number generator has to be specified. This allows the user to repeat simulations with the same series of pseudo-random current injections. White noise current is simulated with pseudo-random numbers generated by a Monte Carlo method. White noise currents cannot be cyclical. Voltage Clamp The command is always enabled if there is a simulation database in memory. It is italicized when the simulation has finished running. A check mark shows whether it is active. Figure V/24: the Voltage Clamp dialog window. Right: voltage clamp of a second neuron. Complex voltage clamps can be simulated, the user only has to specify the length and potential steps for up to 5 clamping periods in one neuron. In each period the potential can be stepped to different values, however if steps (different from 0 mv) are specified for different periods all these periods will step together. Steps are allowed in any period, so there is a free choice of the stepping period within the voltage clamp cycle. Voltage clamps start from the beginning of the simulation and continue for the complete voltage clamp length, i.e. the length of a voltage clamp cycle (the sum of the lengths of all periods) multiplied by the maximum number of steps. Another neuron in the network can be clamped to a continuous, steady potential. To get standard voltage clamp graphics one should select 2 plots (Configure Plots) with repetitive sweeps, time axis from zero to the length of a voltage clamp cycle, and plot the membrane voltage on the first and the clamping current on the second axis. An example is shown in the Test-cell 1 clamp simulation data file. The Voltage Clamp settings can be edited while a simulation is running, but this can cause unexpected effects. Be careful!
V-78 Nodus Menus Delete voltage clamps: clears all Voltage Clamp settings in the simulation database. Voltage Clamp is no longer active. Neuron #1 and compartment selection: The box in the upper part of the dialog window contains either a neuron selection popup menu or a parameter selection popup row (see Selecting Simulation Parameters chapter IV) of the membrane potential type (3 popup menus). If the neuron popup menu shows not used voltage clamps are disabled. All compartments: if this button is selected all compartments in the selected neuron will be clamped to the selected voltages. Only a neuron selection popup menu will be shown in the space above. Compart: if this button is selected only the selected compartment of the selected neuron will be clamped to the selected voltages. Next to this button a compartment preselection popup menu and a compartment selection popup menu allow selection of the clamped compartment. It is up to the user to decide which alternative gives the best approximation of the actual biological experiment. For small, compact cells it might be possible to clamp the complete cell, but in a lot of cases only the soma and the large branches close to the soma are effectively clamped. If one compartment is clamped, then the compartments which are not clamped will still affect the clamping current by passive flow of current to the connected compartments. Periods: For the first neuron 5 clamping periods can be specified, they are executed in the order shown. A clamping period is active if its length is longer than zero. During the first cycle the clamped compartment(s) are set to the from voltage; during the following cycles the clamp voltage is increased by the step value, till the to voltage is reached. The voltage clamp experiment is finished after the cycle during which the period with the largest number of steps has reached its to voltage. Nodus checks if the total simulation time (Computation Settings ) matches the voltage clamp specification, if not then an alert message is shown: Neuron #2 and compartment selection: If neuron #1 is voltage clamped, a steady clamping voltage can be specified for a second neuron. If the neuron popup menu shows not used no second voltage clamp is defined. All compartments: if this button is selected all compartments in the selected neuron #2 will be clamped to the selected voltage. Only a neuron selection popup menu will be shown in the space above. Compart: if this button is selected only the selected compartment of the selected neuron #2 will be clamped to the selected voltage. Next to this button a compartment preselection popup menu and a compartment selection popup menu allow selection of the clamped compartment. clamp at: the second neuron will be clamped at this voltage during the complete voltage clamp experiment. Warning messages When voltage clamp settings have been changed and the user presses OK, the total voltage clamp experiment length is computed and compared with the simulation length. If simulation length is too short a warning is shown: Figure V/25: the warning when simulation length does not match voltage clamp length.
Nodus Menus V-79 No Change: simulation length remains the same, change it manually with Integration Settings. Change Length: the simulation length is reset to the voltage clamp experiment time. Nodus checks whether the settings for the simulation plot (Configure Plots) match the voltage clamp specification, if not then a warning is shown. Figure V/26: the warning when plot settings do not match the voltage clamp settings. No Change: the plot settings remain as they were, change them manually with Configure Plots. Change Plot: the end of all time axes is reset to the length of one clamping cycle, repetitive sweeps is selected and the membrane potential in the first clamped compartment is plotted on the first axis and the clamping current on the second axis. If a second neuron is clamped, a third axis is added to show its clamping current. Synaptic Firing Times The command is always enabled if there is a simulation database containing type 1 synapses in memory. It is italicized when the simulation has finished running. A check mark shows whether it is active. Figure V/27: the Synaptic Firing Times dialog window. This command can be used to specify times at which type 1 synapses will discharge during the simulation in single neuron models or independent of any firing caused by presynaptic neuronal activity in network models. See the Simulation of Synapses and Connections section in chapter III for a definition of a type 1 synapse The Synaptic Firing Times settings can be edited while a simulation is running. This causes no problems, retroactive changes are of course not possible. Delete all firing times: clears all Synaptic Firing Times settings in the simulation database. Synaptic Firing Times is no longer active. Compartment selection: The box in the upper part of the dialog window contains a parameter selection popup row (see Selecting Simulation Parameters chapter IV). These allow selection of a compartment and a type 1 synapse to set firing times form, any changes made to the presently shown firing times will be stored into memory. If no Synaptic Firing Times are defined the first compartment with a synapse in the simulation database will be shown, otherwise the first compartment with preset firing times.
V-80 Nodus Menus Fire single synapses: only the selected synapse will fire (default). Fire all synapses in neuron at once: all the type 1 synapses in all the selected neurons will fire at the preset firing times. This option works on all the preset firing times. The parameter selection popup row will still show compartments and synapses, but only the neuron selection is relevant. Delete all firing times for this synapse: all current injections for the compartment shown are cleared. Current injections in other compartments are not affected. Transmitter amount: (fixed) amount of transmitter that will be released at the preset firing times shown for this synapse. Different transmitter amounts may be specified for different synapses (default is 1.0, arbitrary units). Firing times: For each selected synapse (or neuron) 12 firing times may be specified. They should be entered in consecutive order, from left to right and up to down. Any value entered corresponds to a synaptic firing time (also 0.0 ms). The firing times do not have to be multiples of the major time step (see Integration Settings). Block Ionic Currents The command is always enabled if there is a simulation database containing voltage dependent conductances in memory. It is italicized when the simulation has finished running. A check mark shows whether it is active. Figure V/28: the Block Ionic Currents dialog window. At the right: an example of the special conductance popup menu. This command can be used to selectively block ionic conductances, simulating the bath application of antagonists. Conductances may be blocked completely (100%) or partially (more than 0 %). Several currents may be blocked, but they all have to be blocked during the same period in the simulation. Delete all blocks: clears all blocking factors in the simulation database. Block Ionic Currents is no longer active. Begin: start of the current block period in milliseconds (ms). Should be a be multiple of the major time step (see Integration Settings). End: end of the current block period in milliseconds (ms). Should be a be multiple of the major time step (see Integration Settings). Block: a selection conductance selection popup menu allows the selection of a conductance for which then a blocking factor between 0 and 100% can be specified. The popup menu also shows for each conductance its blocking factor (see Fig. V/28). by: a text box allows entering or editing of the blocking factor for the conductance selected with the conductance selection popup menu. The blocking factor is stored in memory when another Block conductance is selected with the popup menu or OK is pressed. Measure Window The command is enabled if the Plot Window is active. A check mark shows whether a Measure Window is open.
Nodus Menus V-81 This commands opens a Measure Window or selects it and activates measuring (Fig. V/29). Figure V/29: an example of measuring. See the description of the Measure Window at the begin of this section for a description of measuring. Plot Window The command is always enabled if there is a simulation database in memory. A check mark shows whether the Plot Window is shown. This commands shows the Plot Window or selects it. See the description of the Plot Window at the begin of this section. Status Window The command is always enabled if there is a simulation database in memory. A check mark shows whether a Status Window is open. This commands opens a Status Window or selects it and updates the simulation status. See the description of the Status Window at the begin of this section for a description of its contents. Time Window The command is always enabled if there is a simulation database in memory. A check mark shows whether the Time Window is shown. This commands shows the Time Window or selects it. See the description of the Time Window at the begin of this section. Network Menu The network menu is file specific, its commands are enabled only when a network definition window is active (i.e. is in the front) and the commands will operate on the network definition file connected to that window. Only one network definition file can be stored in memory. The New, Open, Close and Save Network commands (File menu) operate on this file. Windows Network definition window A network definition window displays which neurons are part of the network. The name of the corresponding file is the title of the window and is listed at the bottom of the Network menu. A check mark before that menu item shows whether the window is shown, the command can be used to select or open the window.
V-82 Nodus Menus Fig. V/30: a network definition window. The network definition window (Fig. IV/14, V/30) lists all neurons in the network. Each neuron has a local name, and a link to a neuron definition file; lines showing None mean that there is no neuron. Only 8 neurons of up to 40 are shown at a time, they are numbered consecutively (left side of the window). Previous: show the preceding set of 8 neurons, any changes made to the presently shown set are stored in memory. Next: show the next set of 8 neurons, any changes made to the presently shown set are stored in memory. Besides the Next and Previous buttons is space for a Comment describing the network. Local name: the name of the neuron in the network, specific for this network definition. This name will be shown in the neuron popup menu during selection of simulation parameters. If you do not specify a local name, Nodus will generate one if a link to a neuron definition file is set. Neuron definition: a neuron popup menu, listing all neuron definitions in memory, allows selection of a neuron definition file that is linked to the local name of the network neuron. The same neuron definition file can be used for several network neurons (as in Fig. V/30), or the network can consists of different types of neurons (specified by separate neuron definition files). Note that all the neuron definitions linked to a network have to be in memory, or the popup menus will not work properly (see the Open all linked files in the Open Network dialog window). Set to None to remove a neuron from the network. Connections: shows the number of presynaptic contacts (Out) and postsynaptic (In) connections present between this neuron and other neurons in the network. Connections cannot be changed in the network definition window, use the Synaptic Connections menu command. Menu Commands The Network menu commands are active when a network definition window is the highlighted window. At the bottom of the menu all network definition files in memory are listed. Fig. V/31: the Network menu.
Nodus Menus V-83 Synaptic Connections With this command one can specify or edit synaptic connections between the neurons of a network. It is enabled when 2 or more network neurons have been defined. The connection is specified at the presynaptic neuron, each network neuron can be presynaptic to up to 8 other neurons. In Nodus 3.1 all connections are hard-wired, there is no synaptic plasticity implemented. Synaptic connections can have delays, this is a simple but effective way to simulate (long) axons. Fig. V/32: the Synaptic Connections dialog window. Connections from: a neuron popup menu allows the selection of any network neuron as the presynaptic neuron, any changes made to the presently shown connections are stored in memory. The dialog window shows all the connections that have been defined from the selected neuron. Connections Connections are defined by linking a presynaptic compartment (having a transmitter release site) with a specific synapse at a postsynaptic compartment of another neuron. This is done with popup menus. Compartment: a compartment popup menu allows the selection of the presynaptic compartment. As default the first compartment having a transmitter release site in the selected presynaptic neuron will be shown. Neuron: a neuron popup menu allows the selection of the postsynaptic neuron. None is shown for undefined connections (in Fig. V/32 only one connection has been defined, initially all neuron popup menus will be None). One cannot connect a neuron to itself. Compartment: a compartment popup menu allows the selection of the postsynaptic compartment. No popup menu is shown if the postsynaptic neuron is set to None. The compartment popup menu will only show compartments with synaptic currents. As default the first such compartment in the postsynaptic neuron is shown. Synapse: a subdefinition popup menu allows the selection of any of the synaptic current subdefinitions tied to the postsynaptic compartment (up to 3). As default the first synaptic current is shown. Nodus 3.1 does not prevent you form linking transmitter release sites to inappropriate postsynaptic sites (see the Simulation of Synapses and Connections section in chapter III). Delay, ms: the delay between the presynaptic transmitter release and the corresponding postsynaptic conductance in milliseconds (ms). Default is 0 ms, use higher values to simulate axons (as a transmission line delay).
V-84 Nodus Menus Neuron Menu The neuron menu is file specific, its commands are enabled only when a neuron definition window is active (i.e. is in the front) and the commands will operate on the neuron definition file connected to that window. Several neuron definition files can be stored in memory at once. The New, Open, Close and Save Neuron commands (File menu) operate on the file connected to the active window. Windows Neuron definition window A neuron definition window displays general data about the neuron. The name of the corresponding file is the title of the window and is listed at the bottom of the Neuron menu. A check mark before that menu item shows whether the window is shown, the command can be used to select or open the window. Fig. V/33: a neuron definition window. The neuron definition window (Fig. IV/12, V/33) consists of 2 parts, the upper part contains the description and controls for the neuron, the lower part its specific cable parameters. Neuron description: Created on: the day on which the corresponding file was created (with a Save As command). Modified: the day on which the corresponding file was last Saved. Under the creation date is space for a Comment describing the neuron. Number of compartments: the total number of compartments in this neuron definition. Default setting is 100. This number can be changed at any time. Note however that new compartments are not allocated automatically; their sizes and connections have to be defined by the user. If the number of compartments is reduced, Nodus prompts for confirmation before deleting defined compartments. Tree model format: if this option is selected (default) the compartments have to be defined in a centrifugal way, starting from the soma. Nodus checks if the connections follow tree format rules and connects branches automatically back to the parent compartment. See the Nodus Implementation of Compartmental Models section in chapter III for an accurate description of the rules enforced by the tree format. Do not switch off this option unless it is necessary to create an unorthodox model. The tree model format enforces good modeling practices, it also prevents stupid mistakes (which can happen if several 100 compartments have to be defined!). The Tree model format button is on and disabled if the 3-dim coordinates option is on.
Nodus Menus V-85 3-dim coordinates: if this option is selected (default is off) the compartment definition window (Fig. V/35) will show the three dimensional coordinates of the center of the proximal and distal faces of the compartment cylinders. The 3-dim coordinates option can only be used if the Tree model format option is on, because of the way the coordinates are stored in memory. When it is selected the Tree model format button will be disabled. Using 3- dim coordinates has no advantages in Nodus 3.1, it reduces the maximum number of compartments from 4000 to 3000. Future Nodus versions may have 3-dim drawings, which will need these coordinates. Cable parameters The significance of the cable parameters is explained in the The Mathematics of Compartmental Modeling and Passive Membrane Models sections in chapter III. Membrane capacitance: membrane capacitance of the equivalent cable in microfarads per square centimeter (µf/cm 2 ). This value is used for all compartments. Membrane resistance: membrane resistance of the equivalent cable in kilo-ohm-squarecentimeters (kωcm 2 ). This value may vary between compartments: Constant: the value entered for membrane resistance is used in all passive compartments. Variable: membrane resistance can be defined for each passive compartment separately in the compartment definition window (Fig. V/35), the value entered in the neuron definition window is used as default. From compartment # to: this option makes the use of 2 different values for membrane resistance easy. All passive compartments within this range of compartment From to compartment to have for membrane resistance the value specified at the end of the range specification. The first value specified above is used for the other compartments. In most cases the first value will correspond to the membrane resistance for the soma, the second value is used for the dendrites (these can be defined as a continuous range in a tree model format). Cytoplasmic resistance: cytoplasmic resistance of the equivalent cable in ohm-centimeters (Ωcm). Remark that kω is used for membrane resistance and Ω for cytoplasmic resistance! This value is constant for all compartments. Resting membrane potential: resting membrane potential in millivolts (mv) for all passive compartments of the neuron model. The Nodus integration routines perform better if a non-zero value is used; enter a biologically realistic value. Scaling factor: allows the global scaling of all compartment membrane surfaces, without affecting the cytoplasmic resistances between them. See the The Mathematics of Compartmental Modeling section (Eq. 9) in chapter III for more details. The default value is 1.0 (no scaling). Compartment definition window A compartment definition window displays specific data about one compartment. The name of the corresponding neuron definition file, followed by the compartment number, is the title of the window. The name of the file is listed at the bottom of the Neuron menu. A check mark before that menu item shows whether the window is shown, the command can be used to select or open the window. The compartment definition window (Fig. IV/13, V/34, V/36) consists of several parts. The upper part contains the description and morphology of the compartment, below that two lines of text show the compartment cable parameters. A large box at the lower left side shows the connections to other compartments. At the lower right ties to subdefinitions and (if present) the three dimensional coordinates of the compartment are shown.
V-86 Nodus Menus Fig. V/34: a compartment definition window of a constant membrane resistance neuron model without 3-dim coordinates. Compartment description and morphology: Name: name of the compartment. Any word with up to 10 characters can be used to give each compartment a specific name which will be shown in compartment popup menus during selection of simulation parameters. You do not have to give names; by reserving Names for special compartments one can shorten the compartment popup menu if the Show only named comparts option in the Preferences (Edit menu) is on (see the Selecting Simulation Parameters section in chapter IV). Automatic naming is available in the Import Neuron command. Structure type: the structure type popup menu (Fig. V/35) contains 14 predefined structure names that describe different parts of neurons. Use this menu to distinguish between different segments of the neuron. The structure type name will be used in compartment popup menus if no compartment Name is specified (see the Selecting Simulation Parameters section in chapter IV).The structure type has no effect on the compartment shape. Default is soma for the first compartment and undefined for all the following ones. Sphere / Cylinder: select between these 2 options to determine the compartment shape. The compartment shape determines how the membrane surface and cytoplasmic resistance are calculated. Sometimes the Sphere option may be disabled if the tree model option (neuron definition window) is on, because a spherical compartment would be illegal (see the Nodus Implementation of Compartmental Models section in chapter III). Diameter: diameter in micrometer (µm) of the compartment. Has to be larger than zero! Length: length in micrometer (µm) of the compartment. Has to be larger than zero! The Length text box is not shown if the compartment is spherical. Fig. V/35: the structure type popup menu.
Nodus Menus V-87 Compartment cable parameters: The compartment cable parameters are calculated fields (except for Membrane resistance). These values may change if the sizes or shape of the compartment are altered or if some links are added or removed (see the The Mathematics of Compartmental Modeling section in chapter III). The displayed values are updated every time you select another text box or button. Membrane capacitance: the membrane capacitance in picofarad (pf) for the compartment (Fig. V/34). Will only be shown for neuron definitions with a constant value membrane resistance (or 2 values, neuron definition window). Membrane resistance: shows the value for specific membrane resistance in kilo-ohmcentimeter-squared (kωcm 2 ) that is used to calculate the actual membrane resistance of the compartment, shown in kilo-ohm (kω, Fig V/36). The default value for membrane resistance is used until the user has edited the value. Will only be shown for neuron definitions with a variable membrane resistance (neuron definition window). Electrotonic length: the electrotonic length for the compartment. Time constant: the time constant in milliseconds (ms) for a passive compartment. The time constant showed for excitable compartments (tied to a ionic or synaptic currents subdefinitions) is not relevant. Fig. V/36: a compartment definition window of a variable membrane resistance neuron model and with 3-dim coordinates. Connections to other compartments The lower left part of the dialog window contains a list of all the connections or Links to other compartments. For each new connection the compartment to link to, a type of connection (node or branch) and a weight factor (between 1 and 100) is defined. For most compartments this list will start with one or several dimmed connections: the connection to the parent compartment or cross connections (the connections to other proximal compartments for node connections). These connections cannot be changed in this compartment definition, they can only be changed at the definition of the parent compartment, i.e. the more proximal compartment which is (usually) shown as first in the list. The first compartment in each neuron definition can now have up to 24 Links to other compartments, all other compartments have a maximum of 6 links. In the first compartment two buttons control the range of connections displayed, the numbers at the left show the range. >>: moves forward through the connection range (with jumps of 6). <<: moves back through the connection range (with jumps of 6). Link to: the compartment number of the compartment to which this one is connected. If it is proximal relative to the compartment shown, the text box is not editable; distal connections can be edited.
V-88 Nodus Menus The type of connection is selected with an icon (Fig. V/37). Push the desired connection type and the icon will be highlighted (default is a node connection). Dimmed icons refer to connections that have been defined elsewhere (at the parent compartment) and cannot be changed in this compartment definition. If the tree model option is off (neuron definition window) or for spherical compartments all branch connection icons will be dimmed. Fig. V/37: the connection type icons. At the left a node connection icon, at the right a branch connection icon. Node connection: the end of one cylinder is connected to the end of the next cylinder, etc. Each compartment is cross-connected to all other compartments connected by the same node. Branch connection: compartment is connected to the center of the cylindrical compartment of the parent branch. Different branch connections to the same parent compartment are not linked to each other. Weight factor: for weight factors (n) larger than 1, Nodus computes as if n identical compartments are connected to this parent compartment. See the Nodus implementation of compartmental models section in chapter III for a complete definition of node and branch connections and of weight factors. Subdefinitions The lower right part of the compartment definition window contains several subdefinition popup menus that can tie subdefinitions to the compartment. See the Nodus Files section in chapter IV for more details on subdefinitions in neuron definition files. Ion currents: a popup menu listing all ionic currents subdefinitions allows you to add voltage dependent conductances to the compartment, making it excitable. If None is shown the compartment has passive membrane. Synapse #1,2,3: a popup menu listing all synaptic currents subdefinitions allows you to add up to 3 (different) postsynaptic sites to the compartment. No postsynaptic sites are present if all popup menus show None. Transmitter: a popup menu listing all transmitter release subdefinitions allows you to add a presynaptic site to the compartment. There is no transmitter release if the popup menu shows None. Three dimensional coordinates: If the 3-dim coordinates option in the neuron definition window is on, the coordinates will be shown in the lower right corner of the compartment definition window (Fig. V/36). x0, y0, z0: the 3-dim coordinates of the center of the proximal face of the compartment cylinder (or the center coordinates of a spherical compartment). These cannot be edited, they are determined by the coordinates of the parent compartment. x1, y1, z1: the 3-dim coordinates of the center of the distal face of the compartment cylinder (or the center coordinates of a spherical compartment). These can be edited, but be sure that the values fit to the compartment Length! Note also that the 3-dim coordinates are not adapted when the compartment sizes are changed manually. They are adapted when the Optimize Model, Fuse Compartments or Split Compartment commands are used, but not when the Scale Sizes command is used. Menu Commands The Neuron menu commands are active when a neuron definition window or compartment definition window is the highlighted window. At the bottom of the menu all neuron definition files in memory are listed.
Nodus Menus V-89 Fig. V/38: the Neuron menu. Parameters The command is enabled if the front window is a compartment definition window. It removes the present compartment definition window and shows the neuron definition window of the same neuron definition file. Go to Compartment The command is enabled if the front window is a neuron or compartment definition window. Allows the user to display any compartment from the neuron definition file. Fig. V/39: the Go to Compartment dialog window. A dialog window prompts for a compartment number (Fig. V/39). After pressing OK the present neuron or compartment definition window is removed and the compartment definition window of the requested compartment number is shown. If the shift key is pressed when this command is invoked no dialog window will be shown; type the desired compartment number and hit the return key (this is slightly faster). Previous Compartment The command is enabled if the front window is a compartment definition window with compartment number larger than one. The present compartment definition window is removed and the compartment definition window of the preceding compartment (by compartment number) is displayed. Next Compartment The command is enabled if the front window is a neuron or compartment definition window. The present neuron or compartment definition window is removed and the compartment definition window of the next compartment (by compartment number) is displayed. Fuse Compartments The command is enabled if a compartment definition window is active. It can be used to fuse two or more cylindrical compartments linked by node connections together. The total number of compartments will be decremented, other compartments will be renumbered and all branches from the original compartments will be connected to the fused compartment.
V-90 Nodus Menus Fig. V/40: the Fuse Compartments dialog window. The lengths, diameters and 3-dim coordinates of the fused compartment will be computed automatically from the compartment sizes (Eq. 14-17, chapter III). The fused compartment will inherit as many ties to subdefinitions from the original compartments as possible. It will get the name of the first compartment in the range. Use this command to fuse small compartments with the same or similar diameters. Fusing compartments with quite different diameters or fusing many compartments will reduce the electrotonic accuracy of the compartmental model. In the upper part of the dialog window (Fig. V/40) the name of the neuron definition file and the compartment number of the first compartment to fuse is shown. Fuse with compartments to : determines how many compartments will be fused with the one shown at the top. All compartments in between the first (#6 in Fig. V/40) and the last (#7 in Fig. V/40), that are linked by first node connection in the list of connections, will be fused (inclusive the first and the last). The range to fuse should be a continuous, unbranched sequence of compartments, otherwise Nodus will complain. Convert node connections to branches: if this option is selected (default) all sidebranches linked by node connections to compartments between the first and the last of the range to fuse will be converted to branch connections. Because branch connections are linked to the center of the (fused) parent compartment this will slightly improve electrotonic accuracy. If the option is not selected such side-branches will be linked by nodes at the distal side of the fused compartment. This button will be dimmed if no side-branches are present. WARNING: there might not be enough space to connect all side-branches to a fused compartment (maximum of 6 connections). Nodus checks whether there is enough space before changing the connections, but it might run into problems if a lot of compartments are fused and scramble the connections. Better be save than sorry: Save the original file first! Split Compartment The command is enabled if a compartment definition window is active. It splits a cylindrical compartment in two or more compartments with consecutive compartment numbers and linked by node connections. The total number of compartments will be increased and other compartments may be renumbered. Fig. V/41: the Split Compartment dialog window.
Nodus Menus V-91 The lengths and 3-dim coordinates of the split compartment will be computed automatically; all new compartments will have the same length. Ties to ionic currents and synaptic current subdefinitions will be spread over all new compartments if the maximum conductance is determined by compartment surface (i.e. is in ms/cm 2 ), otherwise only the first compartment will keep the ties. Ties to transmitter release sites are always distributed over all the new compartments. They will be renamed automatically. Use this command to split long compartments, it will increase the electrotonic accuracy of the compartmental model (but reduce integration speed). In the upper part of the dialog window (Fig. V/41) the name of the neuron definition file and the compartment number of the compartment to split is shown. Split into compartments: the number of compartments that will replace the original compartment. Default is 2, may be any larger number. Remember however that very short compartments may reduce integration speed enormously! Distribute proximal node connections: some of the side-branches connected by node connections to the proximal part of the original compartment will be reconnected to more distal parts of the split compartment. This distributes the side-branches uniformly over the length of the original compartment. The effect upon electrotonic accuracy depends on the morphology of the real neuron. This button will be dimmed if no proximal side-branches are present. Distribute distal node connections: some of the side-branches connected by node connections to the distal part of the original compartment will be reconnected to more proximal parts of the split compartment. This distributes the side-branches uniformly over the length of the original compartment. The effect upon electrotonic accuracy depends on the morphology of the real neuron. This button will be dimmed if no distal side-branches are present. Convert branch connections to nodes: if this option is selected (default) all sidebranches linked by branch connections to the original compartment will be converted to node connections. This will usually improve electrotonic accuracy. If the option is not selected the branch connections will be distributed over new compartment. This button will be dimmed if no branch connections are present. WARNING: Nodus checks whether there is enough space before changing connections, but it might run into problems in very complex neuron models and scramble the connections. Better be save than sorry: Save the original file first! Optimize Model The command is enabled if the front window is a neuron or compartment definition window. It optimizes the neuron for fast, but accurate integration by fusing or splitting compartments to get compartments with good electrotonic lengths. The total number of compartments may increase or decrease and a lot of compartments will be renumbered. Fig. V/42: the Optimize Model dialog window. The lengths, diameters and 3-dim coordinates of the fused or split compartments will be computed automatically from the compartment sizes (Eq. 14-17, chapter III).
V-92 Nodus Menus This routine calls the Fuse Compartments and Split Compartment routines to do the actual work. See the information about these commands for more details. Use this command after a new neuron model has been made, after making large changes to the cable parameters or after a Scale Sizes command. Optimizing the model should improve both the electrotonic accuracy of the compartmental model and the integration speed. You can control how much the original morphology may be changed. This command can take a long time to execute if there are a lot of compartments! In the upper part of the dialog window (Fig. V/40) the name of the neuron definition file is shown. Electrotonic lengths from to : determines the range of optimal electrotonic lengths for all compartments that Nodus will try to achieve. The suggested values are a good compromise between electrotonic accuracy and integration speed. Fuse compartments if needed: if this option is selected (default), Nodus will fuse short compartments if possible. If it is not selected, electrotonic lengths optimal for fast integration may not be achieved, but the original morphology will be conserved better. Split compartments if needed: if this option is selected (default), Nodus will split long compartments. If it is not selected, electrotonic lengths optimal for accuracy may not be achieved. One should always select this option. Redistribute node connections: if this option is selected some side-branches connected by node connections to a compartment that is split will be reconnected to different parts of the split compartment, to distribute them uniformly over the length of the original compartment. The original morphology will be conserved better if this option is not selected. Branch connections to a split compartment are always converted to node connections. Branch connections may be used: if this option is selected side-branches linked by node connections to compartments that are fused will be converted to branch connections. The original morphology will be conserved better if this option is selected, but the electrotonic accuracy may be diminished. WARNING: there might not be enough space to connect all side-branches to a fused compartment (maximum of 6 connections) or for new compartments. Nodus checks whether there is enough space before changing the connections, but it might run into problems if a lot of compartments are fused and scramble the connections. This command is more dangerous than individual Fuse Compartments or Split Compartment commands. Better be save than sorry: Save the original file first! Scale Sizes The command is enabled if the front window is a neuron or compartment definition window. It irreversibly scales all compartment sizes, without changing the number of compartments or their connections. Fig. V/43: the Scale Sizes dialog window. Use this command to compensate for known global measuring errors in a neuron, for example shrinkage or to change compartment sizes in homogeneous models (e.g. an axon) fast.
Nodus Menus V-93 In the upper part of the dialog window (Fig. V/40) the name of the neuron definition file is shown. The change in sizes are entered as a multiplication factor. Nothing changes for a factor of one (default); sizes decrease for factors smaller than one and increase if it is larger than one. Spherical and cylindrical compartments may be scaled separately. Spheres: diameters multiplied by: scaling factor for spherical compartments. Cylinders: diameters multiplied by: scaling factor for the diameter of cylindrical compartments. lengths multiplied by: scaling factor for the length of cylindrical compartments. Ionic Currents The command is enabled if the front window is a neuron or compartment definition window. It combines ionic current subdefinition formulation and editing functions. See the Nodus Files section in chapter IV for a description of subdefinitions and their relation to neuron definition files. Fig. V/44: the Ionic Currents dialog window. The Ionic Currents dialog window (Fig. V/44, V/45) contains several sections, some are not always functional. At the left side of the dialog window are the data specific for a selected ionic current subdefinition; at the right are general subdefinition controls and info. The OK and Cancel buttons at the right bottom have their usual meaning and close the dialog window, except during the execution of subdefinition management commands (see further). Subdefinition name and data: At the top left of the subdefinition dialog window is a subdefinition selection popup menu labeled as Currents subdefinition, it contains a list of the names of all the ionic current subdefinitions in memory (from all the neuron definition files in memory). The popup menu allows selection of a subdefinition to view or edit, any changes made to the presently shown ion current subdefinition will be stored into memory. If the neuron definition file in the front window uses ionic current subdefinitions, the name of the first one used will be shown; if no ionic current subdefinitions are in memory None will be shown. Under the subdefinition selection popup menu a data box shows the ionic current subdefinition. It can contain up to 10 different currents, each consisting of a link to a conductance definition file, a maximum conductance and a reversal potential (see Eq. 27, chapter III). A current is defined if a conductance name is shown and text boxes for the maximum conductance and the reversal potential are present (lines #1 to 4 in Fig. V/44), else None is shown (lines #5 to 10). Leak stands for a ionic current with a constant conductance (not voltage-dependent, Eq. 24).
V-94 Nodus Menus Conductance: the conductance is selected with a conductance popup menu, which shows the names of all conductance definition files in memory and Leak. Use the popup menu to select a conductance equation for the current, each conductance definition can be selected only once in a ion current subdefinition. Gmax: is the maximum conductance (g-bar in Eq. 24-26). It can be expressed either as a compartment membrane surface dependent value in ms/cm 2 (millisiemens per square centimeter) or an absolute value in ns (nanosiemens), depending on the setting of the conductance options in the Preferences dialog when the ion current subdefinition was created (the units cannot be changed afterwards). Gmax should be larger than zero. Rev. Pot.: the reversal potential (E j in Eq. 27) in millivolts (mv). Used by list: Used by shows a list of all neuron definition files in memory using the selected ionic current subdefinition. This list is important if the Multiple use of subdefinitions check box in the Preferences dialog is switched on; possible interactions of changes to currents with neuron definitions that should not change can be judged. Subdefinition management: Four buttons at the lower right perform general ionic current subdefinition management. They are used to change the subdefinition list as shown in the subdefinition selection popup menu, they do not affect the data content of the subdefinition. During the execution of New, Duplicate or Rename commands all subdefinition management buttons are dimmed and the OK and Cancel buttons at the right bottom have a special meaning; they will not close the dialog window, but confirm or cancel the command being executed. New: creates a new ionic current subdefinition. The contents of the data box are cleared and a name for the new subdefinition is requested (Fig. V/45). Change the Untitled name to something more relevant and press the carriage return key or the OK button to confirm the new name, the data box will show 10 empty current lines (None). Fig. V/45: the Ionic Currents dialog window during a New command. Delete: deletes irreversibly the presently shown ion current subdefinition. If the subdefinition is Used by a neuron definition file a warning will ask for a confirmation of the command. The data are cleared from memory and its name is deleted from the subdefinition selection popup menu. The Ionic Currents dialog will show another ionic current subdefinition or None if no other subdefinitions are in memory. Duplicate: creates a new ionic current subdefinition with the same currents as in the presently shown subdefinition. The contents of the data box are cleared and a name for the new subdefinition is requested (as in Fig. V/45, but the name will be Copy of ). Edit the name and press the carriage return key or the OK button, the data box will return to normal.
Nodus Menus V-95 Rename: renames the presently shown ionic current subdefinition. The contents of the data box are cleared and the old name of the existing subdefinition is shown (as in Fig. V/45, but with the original name). Edit the name and press the carriage return key or the OK button, the data box will return to normal. Transmitter Release The command is enabled if the front window is a neuron or compartment definition window. It combines transmitter release subdefinition formulation and editing functions. See the Nodus Files section in chapter IV for a description of subdefinitions and their relation to neuron definition files. Fig. V/46: the Transmitter Release dialog window. At the right the transmitter data boxes for different types of variable transmitter release are shown. The Transmitter Release dialog window (Fig. V/46, V/47) contains several sections, some are not always functional. At the left side of the dialog window are the data specific for a selected transmitter release subdefinition; at the right are general subdefinition controls and info. The OK and Cancel buttons at the right bottom have their usual meaning and close the dialog window, except during the execution of subdefinition management commands (see further). Subdefinition name and data: At the top left of the subdefinition dialog window is a subdefinition selection popup menu labeled as Transmitter subdefinition, it contains a list of the names of all the transmitter release subdefinitions in memory (from all the neuron definition files in memory). The popup menu allows selection of a subdefinition to view or edit, any changes made to the presently shown transmitter release subdefinition will be stored into memory. If the neuron definition file in the front window uses transmitter release subdefinitions, the name of the first one used will be shown; if no transmitter release subdefinitions are in memory None will be shown. Under the subdefinition selection popup menu a data box shows the transmitter release subdefinition. The fields shown depend on the type of transmitter release selected (Fig. V/46). Potential threshold: transmitter will be released whenever the membrane voltage of the compartment to which this subdefinition is tied, is larger than or equal to the potential threshold (E th in Eq. 29-31, chapter III) in millivolts (mv). If the membrane voltage is below the potential threshold, transmitter release is zero. This value is very important for constant transmitter release, because it determines when the type 1 synapse connected to it will fire. For variable transmitter release it can improve integration speed; below this value no transmitter release will be computed (otherwise Nodus might compute a zero or infinitely small release every minor time step). Base amount: fixed amount of transmitter released (constant release) or minimum amount of transmitter released (variable release). The base amount (b) is further defined in Eq. 29-31, chapter III.
V-96 Nodus Menus Constant transmitter release: select this option to have one pulse of constant transmitter release every time the membrane voltage crosses the potential threshold. Connect to a type 1 synapse. Variable transmitter release: Linear to presynaptic potential: select this option to have a simple model of variable, graded transmitter release (Eq. 29). Connect to a type 2 synapse. Factor: difference between membrane voltage and potential threshold (in mv) is scaled by this value (f in Eq. 29) to get the amount of transmitter released at every minor time step. Should be larger than zero. Exponential to presynaptic potential: select this option to have a more complex model of variable, graded transmitter release (Eq. 30). Connect to a type 2 synapse. Factor: the exponential is scaled by this value (f in Eq. 30) to get the amount of transmitter released at every minor time step. Should be larger than zero. Characteristic potential: difference between membrane voltage and potential threshold (in mv) is divided by this value (c in Eq. 30) before the exponential is computed. Should not be zero. Process dependent: select this option to supply your own model of variable, graded transmitter release (Eq. 31). Not implemented in Nodus 3.1. Used by list: Used by shows a list of all neuron definition files in memory using the selected transmitter release subdefinition. This list is important if the Multiple use of subdefinitions check box in the Preferences dialog is switched on; possible interactions of changes to transmitter equations with neuron definitions that should not change can be judged. Subdefinition management: Four buttons at the lower right perform general transmitter release subdefinition management. They are used to change the subdefinition list as shown in the subdefinition selection popup menu, they do not affect the data content of the subdefinition. During the execution of New, Duplicate or Rename commands all subdefinition management buttons are dimmed and the OK and Cancel buttons at the right bottom have a special meaning; they will not close the dialog window, but confirm or cancel the command being executed. New: creates a new transmitter release subdefinition. The contents of the data box are cleared, buttons are dimmed and a name for the new subdefinition is requested (as in Fig. V/47, but the name will be Untitled). Change the Untitled name to something more relevant and press the carriage return key or the OK button to confirm the new name, the data box will return to normal. Fig. V/47: the Transmitter Release dialog window during a Duplicate command.
Nodus Menus V-97 Delete: deletes irreversibly the presently shown transmitter release subdefinition. If the subdefinition is Used by a neuron definition file a warning will ask for a confirmation of the command. The data are cleared from memory and its name is deleted from the subdefinition selection popup menu. The Transmitter Release dialog will show another transmitter release subdefinition or None if no other subdefinitions are in memory. Duplicate: creates a new transmitter release subdefinition with the same transmitter equations as in the presently shown subdefinition. The contents of the data box are cleared, buttons are dimmed and a name for the new subdefinition is requested (Fig. V/47). Edit the Copy of name and press the carriage return key or the OK button, the data box will return to normal. Rename: renames the presently shown transmitter release subdefinition. The contents of the data box are cleared, buttons are dimmed and the old name of the existing subdefinition is shown (as in Fig. V/47, but with the original name). Edit the name and press the carriage return key or the OK button, the data box will return to normal. Synaptic Currents The command is enabled if the front window is a neuron or compartment definition window. Fig. V/48: the Synaptic Currents dialog window. At the right the synaptic conductance data boxes for different types of synaptic currents are shown. The Synaptic Currents dialog window (Fig. V/48-V/50) contains several sections, some are not always functional. At the left side of the dialog window are the data specific for a selected synaptic current subdefinition; at the right are general subdefinition controls and info. The OK and Cancel buttons at the right bottom have their usual meaning and close the dialog window, except during the execution of subdefinition management commands (see further). Subdefinition name and data: At the top left of the subdefinition dialog window is a subdefinition selection popup menu labeled as Synaptic subdefinition, it contains a list of the names of all the synaptic current subdefinitions in memory (from all the neuron definition files in memory). The popup menu allows selection of a subdefinition to view or edit, any changes made to the presently shown synaptic current subdefinition will be stored into memory. If the neuron definition file in the front window uses synaptic current subdefinitions, the name of the first one used will be shown; if no synaptic current subdefinitions are in memory None will be shown. Under the subdefinition selection popup menu a data box shows the synaptic current subdefinition. The fields shown depend on the type of transmitter release selected (Fig. V/48). Reversal potential: reversal potential of the synaptic current (E s in Eq. 37, chapter III) in millivolts (mv).
V-98 Nodus Menus Peak conductance: is the maximum conductance (g-bar in Eq. 32-36). It can be expressed either as a compartment membrane surface dependent value in ms/cm 2 (millisiemens per square centimeter) or an absolute value in ns (nanosiemens), depending on the setting of the conductance options in the Preferences dialog when the synaptic current subdefinition was created (the units cannot be changed later). It should be larger than zero. Synaptic conductance: Five different types of synaptic conductance equations are available. These determine whether the subdefinition is a type 1 or a type 2 synapse (see the Simulation of Synapses and Connections section in chapter III). Constant: select this option to let the synaptic conductance always be equal to the Peak conductance. Is a type 2 synapse (Eq. 32). Alpha function: select this option for a standard, simple model of a variable synaptic conductance. Is a type 1 synapse (Eq. 33). Time to peak: determines time course of the synaptic conductance (τ in Eq. 33). Unit is milliseconds (ms), should be larger than zero. Factor alpha: determines sharpness of change in synaptic conductance (α in Eq. 33). Should be one or larger (integer value). Dual exponential function: select this option for a simple model of a variable synaptic conductance. Is a type 1 synapse (Eq. 34). Open time constant: determines initial time course of the synaptic conductance (τ o in Eq. 34). Unit is milliseconds (ms), should be larger than zero. Close time constant: determines final time course of the synaptic conductance (τ c in Eq. 34). Unit is milliseconds (ms), should be larger than zero and different from the Open time constant. Conductance dependent: select this option to use a voltage dependent conductance equation for the synaptic conductance (Eq. 35). Conductance: choose a conductance definition file with the conductance popup menu. Is a type 2 synapse. Process dependent: select this option to supply your own model of variable synaptic conductance (Eq. 36). Not implemented in Nodus 3.1. Plot time course: this button is enabled when a type 1 synaptic conductance is selected. A small diagram showing the time course of the synaptic conductance, normalized to the range 0 to 1, is drawn in the Used by box, which is renamed to Conductance plot (Fig. V/49). One (alpha function) or 2 red bars (dual exponential function) at the bottom show the Time to peak or Open and Close time constants. At the top right the time to peak (in ms) is shown. Several Conductance plots can be superimposed.
Nodus Menus V-99 Fig. V/49: the Synaptic Currents dialog window after a Plot time course command. Used by list: Used by shows a list of all neuron definition files in memory using the selected synaptic current subdefinition. This list is important if the Multiple use of subdefinitions check box in the Preferences dialog is switched on; possible interactions of changes to synaptic conductance equations with neuron definitions that should not change can be judged. The Used by box is replaced by a Conductance plot after the Plot time course button is pressed (see above). The Used by box will be shown again if another synaptic current subdefinition is selected. Subdefinition management: Four buttons at the lower right perform general synaptic current subdefinition management. They are used to change the subdefinition list as shown in the subdefinition selection popup menu, they do not affect the data content of the subdefinition. During the execution of New, Duplicate or Rename commands all subdefinition management buttons are dimmed and the OK and Cancel buttons at the right bottom have a special meaning; they will not close the dialog window, but confirm or cancel the command being executed. New: creates a new synaptic current subdefinition. The contents of the data box are cleared, buttons are dimmed and a name for the new subdefinition is requested (as in Fig. V/50, but the name will be Untitled). Change the Untitled name to something more relevant and press the carriage return key or the OK button to confirm the new name, the data box will return to normal. Delete: deletes irreversibly the presently shown synaptic current subdefinition. If the subdefinition is Used by a neuron definition file a warning will ask for a confirmation of the command. The data are cleared from memory and its name is deleted from the subdefinition selection popup menu. The Synaptic Currents dialog will show another synaptic current subdefinition or None if no other subdefinitions are in memory. Duplicate: creates a new synaptic current subdefinition with the same conductance equations as in the presently shown subdefinition. The contents of the data box are cleared, buttons are dimmed and a name for the new subdefinition is requested (as in Fig. V/50). Edit the Copy of name and press the carriage return key or the OK button, the data box will return to normal. Rename: renames the presently shown synaptic current subdefinition. The contents of the data box are cleared, buttons are dimmed and the old name of the existing subdefinition is shown (Fig. V/50). Edit the name and press the carriage return key or the OK button, the data box will return to normal.
V-100 Nodus Menus Fig. V/50: the Synaptic Currents dialog window during a Rename command.
Nodus Menus V-101 Conductance Menu Windows Conductance definition window A conductance definition window shows the conductance equations. All ion conductances are defined by Hodgkin-Huxley like equations, see the Excitable Membrane Models section in chapter III for more information. The name of the corresponding file is the title of the window and is listed at the bottom of the Conductance menu. A check mark before that menu item shows whether the window is shown, the command can be used to select or open the window. Fig. V/51: the conductance definition window. Initially the activation (M) equations are shown, press on the H button to see the inactivation rate factors (right). The conductance definition window (Fig. IV/11, V/51) shows at the top the general conductance equation which can be of 2 forms: 2 states: the ion channel can be closed or open; this is controlled by the activation factor M (Eq. 25 in chapter III). 3 states: the ion channel can be closed, open, or inactivated; this is controlled by the activation factor M and the inactivation factor H (Eq. 26). Both factors may be raised to a power (4 for activation and 1 for inactivation in Fig. V/51). Below the general equation the (in)activation factor equation (Eq. 21) and the rate factor equations (Eq. 20) are shown. Only one set of equations can be shown at a time, either activation or inactivation. Press the M-button to see the activation equations or the H-button to see the inactivation equations (Fig. V/51). The rate factor equations and the power values above M and H are text boxes so that the values can be changed to get equations specific for different conductances. Under the equations is space for a Comment describing the conductance. Conductance Plot window Fig. V/52: a Conductance Plot window showing activation (M) and inactivation (N) versus voltage.
V-102 Nodus Menus The Plot command can make several types of Conductance Plot windows (Fig. V/52). These windows are graphics windows and do not activate the Conductance menu. The name of each trace in the Conductance Plot window is shown in the correct color in the upper right corner. On black and white monitors the traces will be drawn with different patterns. Menu Commands The Conductance menu commands are active when a conductance definition window is the highlighted window. At the bottom of the menu all conductance definition files in memory are listed. Fig. V/53: the Conductance menu. Plot Scales The command is enabled if the front window is a conductance definition window. It sets the ranges for the potential and value axes of all Conductance Plot windows created by subsequent Plot commands. Fig. V/54: the Plot Scales dialog window with automatic (right) or manual (left) control of the value axis range (right). Potential range: range for the potential axis (X-axis) of new Conductance Plot windows. In millivolts (mv), default is -80 to +40 mv. Value range: range for the value axis (Y-axis) of new Conductance Plot windows. Default is automatic; the Plot command will compute the maximum value in the selected Potential range and draw an value axis from zero to the maximum value. If the automatic button is not selected, 2 text boxes ( to U, Fig. V/54 right) are shown and the user can set other ranges for the value axis. Use this manual option to enlarge parts of the plot. Overlay existing plots: if a Conductance Plot window of the active conductance and of the selected value type already exists, the Plot command will draw a new plot in the existing window. Useful to visualize the effect of changes in the conductance equations. Default is off, each Plot command will create a new window. The Overlay option is not implemented in Nodus 3.1. Plot Conductance Plots the steady state conductance (G ) versus membrane potential in a Conductance Plot window. The axis ranges shown depend on the setting of the Plot Scales command.
Nodus Menus V-103 Plot (In)Activation Plots the steady state activation factor (M ) and (if present) the steady state inactivation factor (H ) versus membrane potential in a Conductance Plot window. The axis ranges shown depend on the setting of the Plot Scales command. Plot Time Constants Plots the activation time constant (τ M ) and (if present) the inactivation time constant (τ H ) versus membrane potential in a conductance plot window. The axis ranges shown depend on the setting of the Plot Scales command. Plot Rate Factors Plots the activation rate factors (a m and b m ) and (if present) the inactivation rate factors (a h and b h ) versus membrane potential in a Conductance Plot window. The axis ranges shown depend on the setting of the Plot Scales command.
V-104 Nodus Menus
Nodus Examples VI-105 VI. USING THE EXAMPLES The Nodus Master disk contains several example files to demonstrate how to use Nodus 3. This chapter describes the content and use of these files. Demo Files Demo files are based on extremely simple models, which do not attempt to be realistic. Examine these files to learn more about using Nodus 3. Suggested user actions have been underlined. Only the output of the simulations is shown here, refer to chapter V for figures showing the menu command dialog windows. Some files show the speed and accuracy of integration. Others demonstrate basic biophysical processes, for example saturation of synaptic current. The demo files are described in an order optimal for discovering Nodus. Their names reflect a different, older sequence of numbering which has been retained for compatibility with Nodus 2. Test-cell 2 Demo Test-cell 2 Demo is a simulation data file, it contains data for the simulation of a simple experiment in a 5-compartment model with 3 voltage dependent ion currents in one of the compartments. Select the Open Simulation command in the File menu, open the Nodus 3 Data folder on the Nodus Master disk and double-click on the Test-cell 2 Demo. Nodus loads the simulation database and the linked neuron definition file ( Test-cell 2 ) and conductance definition files ( CS Fast Na Current, CS Delayed Rectifier and CS A Current ). These file names are listed at the bottom of the respective file menus, press on the Neuron menu title to look at the bottom of the menu and then release, repeat with the Conductance menu. Two simulation windows are created: a simulation plot window, titled Test-cell 2 Demo.Sim0000 and a time window (Fig. VI/1). Because a simulation window is active most of the Simulation menu commands are enabled, press on the Simulation menu title to look at the menu and then release. Fig. VI/1: the simulation plot and time windows of the Test-cell 2 Demo simulation. The simulation Plot Window title consists of the file name, followed by.sim and the simulation number (it may be larger than zero if Nodus has already been used) The window has been configured to show three plots: the upper axis shows the membrane voltage in compartment #1 (the soma of Test-cell 2), the middle axis shows the ion currents in compartment #2 (the spike initiation zone): the CS Fast Na Current, the CS Delayed Rectifier and the CS A Current, the lower axis shows the current injected into the soma (compartment #1).
VI-106 Nodus Examples Select the Run command in the Simulation menu to start the simulation. After a slight delay (used by the hybrid integration method to generate an equations table for the conductance factors), the simulation will start (Fig. VI/2). Initially a small current is injected to make the cell fire, after 600 ms a ramp current with a period of 200 ms is injected. The initial hyperpolarization during the ramps removes the inhibition of the A-current, which is much increased during the spike in the depolarization phase of the ramp (middle axis, green trace). Let it Run till it finishes after 1000 milliseconds (ms) of simulated action potentials. mv 20 0-20 -40 ms 250 500 750 1000 na 2 0-2 -4 na 0.5 0.0-0.5 ms 250 500 750 1000 ms 250 500 750 1000 Fig. VI/2: the plot output of the Test-cell 2 Demo simulation after a complete Run. This demo demonstrates several features of Nodus: voltage dependent ion currents, sophisticated graphic output, complex current injections, etc. We will now examine how the simulation was configured. One can view the different settings in the Simulation menu, but not change them (because this simulation has finished running). This is shown in the Simulation menu by italic menu commands, press on the Simulation menu title to look at the menu and then release. To be able to change the settings a new simulation database has to be created. Select the New Simulation command in the File menu, do not change any settings (Fig. V/3, Use Test-cell 2 Demo for New simulation database and Resting potential for Initial values) and press OK. A new simulation plot window ( Test-cell 2 Demo.Sim2 ) is created. Select the Measure command in the Simulation menu (Fig. V/29). Measure the peak amplitude of the A-current (green trace in the second plot) during the 4 action potentials. Put the hair cursor on the top of the action potential (first plot), press the mouse button and move to the second action potential to measure the period of firing. Close the Measure window to finish. Select the Integration Settings command in the Simulation menu (Fig. V/20). Examine the Time controls and Integration method settings, change them if you like to, and press OK or Cancel if you want to undo any changes. Select the Configure Plots command in the Simulation menu (Fig. V/22). Examine the settings for the different axes by pressing the Edit axis # 1, 2 or 3 buttons. Press on the popup menus to see how values to plot are selected, in particular the compartment popup menus (the third from the left) of axis 1 and 3 and the subdefinition popup menus (fourth from the left) of axis 2. Press OK when you have finished.
Nodus Examples VI-107 Suggested exercises: add plots for membrane potential in all compartments (Edit axis #1, change compartment popup menu in first popup row to #1 Soma, press on the neuron popup menu (first at left) of the second row and change from not used to Test-cell 2, change the compartment popup menu to #2 main segment, repeat for rows 3 to 5 and set to compartments #3 to #5); add a fourth axis showing A-current activation and inactivation (add an axis by pressing on the fifth icon above the Edit axis # row and Edit axis #4, change the value axis popup menu to conductance (in)activation with a range from 0.0 to 1.0 U, select Test-cell 2 on the first popup row neuron popup menu and set the subdefinition menu to #1 CS A Current:M, repeat for second popup row and set to #1 CS A Current:H, note that there is not enough space to show the complete titles of the popup menus). You can Run the simulation to watch the effect of the new Configure Plot settings; make a New Simulation again after it has finished running. Reset to the original Test-cell 2 Demo simulation database. If the present simulation has not finished running you will have to select the Close command in the File menu (you might want to press on the option key, this will conserve the graphics in the Plot Window). Open Simulation the Test-cell 2 Demo file again. Select the Current Clamp command in the Simulation menu (Fig. V/25). Examine the settings for current injections in the Soma of Test-cell 2 by pressing on the Next and Previous buttons. Press OK when you have finished. Suggested exercises: remove the ramp currents and change the amplitude of the steady current to see how it affects spiking frequency (press Next till Current #3 is shown and press Delete, press Delete again for Current #2, change End of Current #1 to 1000 ms and change Amplitude and press OK, Run Simulation, New Simulation and repeat for another Amplitude of Current #1, examine frequencies for amplitudes ranging from 0 to 1.0 na); change steady current to current pulses (after preceding exercise, set Current #1 to Cyclical, set Period to 250 ms and On from 0 to 50 ms, press OK and Run Simulation, check effect of changing amplitude). Test-cell 2 Demo is also used to demonstrate the speed and accuracy of the different integration methods available in Nodus. The simulation computation times (in minutes) were measured with the Status Window command on a Macintosh II (High multifinder priority in Preferences menu on). Integration method Time Rate factor table 0.20 Hybrid Euler, V=0.50 mv 1.32 Hybrid Euler, V=0.10 mv 2.57 Hybrid Euler, V=0.01 mv 18.95 Fehlberg, Rel. Err=10-4 2.75 Fehlberg, Rel. Err=10-6 3.43 Fehlberg, Rel. Err=10-8 5.73 It makes little sense to set V max in the hybrid Euler method smaller than 0.1; it will compute much slower and still less accurate than the Fehlberg method with a relative error of 10-4. The accuracy of the different integration methods was tested with the maxima option in the Text Output command (table on next page). Remark that the largest inaccuracy in the hybrid method occurs in the firing period, but the values are equivalent to the results obtained with the Fehlberg method. The size of the overshoot (and of the afterhypolarization) are less sensitive to the integration method.
VI-108 Nodus Examples Integration method Action potential #1 Action potential #2 Firing Time (ms) Size (mv) Time (ms) Size (mv) period (ms) Hybrid Euler, V=0.50 mv 19.571 36.648 612.584 36.313 593.013 Hybrid Euler, V=0.10 mv 19.665 36.620 613.553 36.296 593.888 Hybrid Euler, V=0.01 mv 19.721 36.622 615.231 36.314 595.510 Fehlberg, Rel. Err=10-4 19.745 36.604 616.682 36.322 596.937 Fehlberg, Rel. Err=10-6 19.747 36.604 616.710 36.322 596.963 Fehlberg, Rel. Err=10-8 19.728 36.604 616.704 36.322 596.976 Test-cell 2 Test-cell 2 is the neuron definition file that contains the description of the simple invertebrate neuron model used in Test-cell 2 Demo. It is a 5-compartment model (corresponding to 8 compartments when the weight factors are expanded, Fig. VI/3) with 3 voltage dependent ion currents in one of the compartments (the spike initiation zone). This model is an example of the use of ionic currents, of node connections and branch connections and of weight factors in a neuron model definition, it is not a good model for a real neuron. 4 5 1 2 3 4 5 5 Fig. VI/3: diagram of Test-cell 2. The compartment numbers are shown, compartment #2 has ionic currents. If you have Open Simulation the Test-cell 2 Demo the Test-cell 2 neuron definition file is already in memory, select the corresponding menu item at the bottom of the Neuron menu, otherwise use the Open Neuron command in the File menu to open Test-cell 2. The neuron definition dialog window (Fig. V/33) shows the number of compartments and the cable parameters used in this model. Select the Next Compartment command in the Neuron menu to look at the Soma compartment (#1). A compartment definition dialog window (Fig. V/34) is shown, look at the description of the spherical morphology (top). The connection to compartment #2 is shown in the box at the bottom left; this is a parent compartment, the link is not dimmed. Select the Next Compartment command again to look at compartment #2 (main segment). This is a cylindrical compartment (top right). Look at the connections (bottom left); the first is dimmed because this is the child compartment for the connection to the Soma; other links are enabled because compartment #2 is the parent to 4 other connections in the model; note that the first 2 are node connections with no weight factors (1) and the last 2 are branch connections with weight factors. Compartment #2 is excitable; it has a tie to an ionic currents subdefinition, shown by the Connor Stevens Currs popup menu title at left. Use the Ionic Currents command in the Neuron menu to look at the subdefinition, press Cancel when done.
Nodus Examples VI-109 Look at one of the conductance definition files used by this ionic currents subdefinition. Select the CS A Current conductance definition window (Fig. V/51) by choosing the corresponding menu item in the Conductance menu. Look at the activation factor equations, press H to look at the inactivation factor equations. Select the Plot (In)Activation command in the Conductance menu to get a Conductance Plot window. Select the Test-cell 2 compartment definition window (it is listed at the bottom of the Neuron menu) and look at the other 3 compartments with the Next Compartment command. Suggested exercises examine how changing the model affects the Test-cell 2 Demo simulation. Be sure to have an old (Closed) simulation database in memory (see description of that demo). Change the maximum conductances of the ionic currents to see how spiking threshold and frequency is affected (select the Test-cell 2 neuron or compartment definition window, select the Ionic Currents command and change one or several Gmax values, press OK, select the New Simulation command, click on the compile form Test-cell 2 button in the New simulation database section, click on the Values in memory button in the Initial values section, press OK, Run the simulation); increase the length of the axon and see how spiking threshold and frequency is affected (select the Test-cell 2 neuron definition window with the Parameters command in the Neuron menu, increase the number of compartments to 6, select the Go to Compartment # command in the neuron menu, type 3 and press OK, add a connection to compartment 6 by typing a 6 in the first text box on the second line of the connections list, Go to Compartment # 6 and type a Diameter of 5 µm and a Length of 300 µm, make a New Simulation and Run it); change a conductance equation (select the conductance equation window, press on the H and change the 56.00 number in the am equation to 46.00,make a New Simulation and Run it). Test-cell 6 Demo 1 This simulation run file demonstrates temporal summation of two IPSPs, shown at all 5 compartments in a simple neuron model (Fig VI/4). mv -40 ms 15 30 45 60-50 -60-70 -80-90 ns 180 160 140 120 100 80 60 40 20 0 ms 14 20 na 1.5 1.0 0.5 0.0-0.5-1.0-1.5 ms 14 20 Fig. VI/4: the plot output of the Test-cell 6 Demo 1 simulation after a complete Run.
VI-110 Nodus Examples Two IPSPs are fired in compartment 5. The upper plot shows the membrane potentials, the lower left plot synaptic conductances and the lower right plot synaptic currents. Note that in all 3 plots traces with the same color refer to the same compartment. The synaptic current saturates during the second IPSP, while the conductances are for the two IPSPs are identical. Open Simulation Test-cell 6 Demo 1 (this is possible only if the previous simulation has finished Running or is Closed) and Run it. Do New Simulation and examine the Integration Settings and Configure Plots dialog windows. Examine the Synaptic Firing Times command dialog window (Simulation menu, Fig V/27) to see how the 2 IPSPs are fired. Suggested exercises: fire a third IPSP (select the Synaptic Firing Times command, type 18 in the third Firing times edit box, press OK, do New Simulation (with Use Test-cell 6 Demo 1 for New simulation database) and Run it); fire IPSPs in another compartment (select the Synaptic Firing Times command, press the Delete all firing times button, select another compartment and synapse with the compartment popup menu (third from the left), type in the Firing times, press OK, do New Simulation and Run it); examine spatial summation by firing synapses in different compartments together. Test-cell 6 This neuron definition file contains a small 5 passive membrane compartment model that is used in to demonstrate synaptic currents. The model and the demonstrations in Test-cell 6 Demo 1 are based on Fig. 2 of Perkel, D.H., Mulloney, B., and Budelli, R.W.: Quantitative methods for predicting neuronal behavior. Neuroscience 4 (1981) 823-837. If you have Open Simulation the Test-cell 6 Demo 1 the Test-cell 6 neuron definition file is already in memory, select the corresponding menu item at the bottom of the Neuron menu, otherwise use the Open Neuron command. A neuron definition dialog window is made. Each compartment has one synapse: #1 to #4 are excitatory ( Fast EPSP and Slow EPSP ), #5 is inhibitory ( Slow IPSP ). Select the Synaptic Currents command (Neuron menu) to look at the synaptic currents subdefinitions (Fig V/48). Only one subdefinition is shown, use the Synapse subdefinition popup menu at the top left to see another subdefinition and the Plot time course button to see a fast graphic display of the synaptic conductance, press Cancel when done. Suggested exercises: change the time constants of a synaptic conductance (Synaptic Currents command, select the Slow IPSP Synapse subdefinition, press Plot time course, change the close time constant to 1.2, press Plot time course, press OK, do New Simulation (with Compile from Test-cell 6 1 for New simulation database) and Run it); add voltage dependent ionic currents and see whether a spike is triggered after an EPSP (Go to Compartment #1, (the Test-cell 2 file must be in memory), do Ionic Currents and select the Connor Stevens Currs (if not shown), press Duplicate and OK, select the Copy of Connor Stevens Currs in the ion currents popup menu of the compartment definition window of Test-cell 6,do New Simulation (with Compile from Test-cell 6 1 for New simulation database) and Run it). Test 7 Network Demo This simulation run file shows reciprocal inhibition between 2 neurons in a small network (Fig VI/5). The difference in firing frequency is controlled by current injection (Current Clamp). Open Simulation Test 7 Network Demo (this is possible only if the previous simulation has finished Running or is Closed) and Run it. Do New Simulation and examine the Configure Plots and Current Clamp dialog windows; remark the use of the neuron definition popup menus (the first one in the row) to distinguish between the 2 neurons in the network.
Nodus Examples VI-111 Test 7 Network This network definition file models an extremely simple network of 2 small neurons. If you have Open Simulation the Test 7 Network Demo the Test 7 Network network definition file is already in memory, select the corresponding menu item at the bottom of the Network menu, otherwise use the Open Network command in the File menu to open Test 7 Network. The network definition dialog window (Fig. V/30) is shown. mv 20 0-20 1.25 2.50 3.75 5.00 sec -40-60 mv 20 0-20 1.25 2.50 3.75 5.00 sec -40-60 Fig. VI/5: the plot output of the Test 7 Network Demo simulation after a complete Run. Examine the synaptic connections between the 2 neurons with the Synaptic Connections command in the Network menu (Fig. V/32). Use the Connections from popup menu to look at the connections from the Left cell and from the Right cell. Suggested exercises: change the (axonal) delay between the 2 neurons (Synaptic Connections command, change the Delay from the Left cell to Right cell to 50 ms, press OK, do New Simulation (with Compile from Test 7 Network for New simulation database) and Run it); add a third neuron to the network (select the network definition window, type Middle cell for Local name in the third row and select Test-cell 7 in the neuron definition popup menu, use the Synaptic Connections command to connect both the Left cell and Right cell to the Middle cell by changing the Neuron popup menu in the second row from None to Middle cell and press OK, do New Simulation and Run it). Test-cell 7 This neuron definition file is similar to the Test-cell 2 model, with a synapse added to the compartment #4 dendrite and a transmitter release site to the compartment #5 dendrite. Use the Transmitter Release and Synaptic Currents commands in the Neuron menu to examine the equations used, both the dialog windows are very similar.
VI-112 Nodus Examples Test-cell 1 Clamp This simulation data file illustrates the use of voltage clamps. It reproduces the experiment that shows the existence of an A-current: a long hyperpolarization followed by a small depolarization (as in Fig. 1B of Connor JA and Stevens CF: Voltage clamp studies of a transient outward membrane current in gastropod neural somata. J. Physiol. (London), 213 (1971) 21-30.). mv -40-50 -60-70 -80-90 na 100 80 60 40 20 0 U 0.8 0.6 0.4 0.2 0.0 ms 500 1000 1500 2000 ms 500 1000 1500 2000 ms 500 1000 1500 2000 Fig. VI/6: the plot output of the Test-cell 1 Clamp simulation after a complete Run. The upper axis shows the membrane voltage in the unique compartment, the middle axis the clamping current and the lower axis activation and inactivation factors for the A-current ( CS A Current ). The mechanism of A-current activation is demonstrated: a long hyperpolarization slowly removes inactivation, depolarization instantly raises the activation factor. Use the Voltage Clamp command Configure Plots in the Simulation menu to examine how this simulation has been constructed. Test-cell 1 This neuron definition file contains the model that Connor and Stevens used to simulate the ion currents in an Anisidoris neuron (Connor JA and Stevens CF: Prediction of repetitive firing behaviour from voltage clamp data on isolated neurone soma. J. Physiol. (London), 213 (1971) 31-53.). It is extremely simple: one excitable compartment, therefore it is only useful for ion current simulations. The ion conductance equations are contained in the 3 CS Current conductance definition files. Test-cell 3 This neuron model was used to test the accuracy of the Nodus integration methods. The 27 compartment model (with many weight factors) is equivalent to a linear cable model of a spinal α-motoneuron for which the analytical results are known (Rall, W: Branching dendritic trees and motoneuron membrane resistivity. Exp. Neurol., 1 (1959) 491-527; Segev, I, Fleshman, JW, Miller, JP and Bunow, B: Modeling the electrical behavior of anatomically complex neurons using a network analysis program: passive membrane. Biol. Cybern., 53 (1985) 27-40.).
Nodus Examples VI-113 The input resistance (R N ) of the model was measured with a hyperpolarizing current injection, the time constant (τ m ) and electrical length (L) were calculated from exponential peeling data (Nodus 2). Integration method R N τ m L MΩ % Err ms % Err %Err Analytical 1.57 7.00 1.46 Hybrid Euler, V=0.1 mv 1.51-4 7.53 +8 1.58 +8 Fehlberg, Rel. Err=10-4 1.51-4 7.09 +1 1.51 +3 Fehlberg, Rel. Err=10-8 1.51-4 6.98 +0 1.52 +3 The results of this accuracy evaluation compare favorably with other compartmental simulation programs (Segev et al, 1985). Test-cell 4a and 4b Test-cell 4a and Test-cell 4b are two different models of the same (extremely simple) neuron. Test-cell 4a uses only node connections, while in Test-cell 4b compartments have been fused and some compartments have branch connections (8 compartments instead of 13). These models were used to check the accuracy of branching versus node connections. A -1.0 na, 1000 ms current was injected in the soma (compartment #1, Nodus 2, double precision Fehlberg method, relative error was 10-8) : Model V (mv) in compartment # 1 4 5 8-10/7 4a -45.62-44.38-43.99-43.23 4b -45.71-44.25-43.86-42.85 Error (%) -0.20 +0.29 +0.30 +0.88 The errors caused by the use of branch connections are minimal, both in the soma and in the branch compartments! A -0.5 na, 1000 ms current was injected in a branch (compartment #5, Nodus 2, double precision Fehlberg method, relative error is 10-8) : Model V (mv) in compartment # 1 4 5 8-10/7 4a -21.99-25.17-44.67-22.42 4b -21.93-26.99-46.47 22.34 Error (%) +0.27-7.23-4.03 +0.36 There is a significant error in all compartments of the side-branch in which current was injected, but in the other compartments, including the soma, the error is minimal. This example shows that a branch connection is sufficiently equivalent to a node connection at the center of the (split) parent compartment for most experiments. Test-cell 5a, 5b and 5c Test-cell 5a, Test-cell 5b and Test-cell 5c are neuron definition files containing three different models of the same example neuron: 5a is the original model and 5b and 5c correspond to the two reduction steps described in the Nodus Implementation of Compartmental Models section in chapter III, Fig III/6. The loss of accuracy caused by the model reduction was estimated by a -0.5 na, 400 ms current in the soma (compartment #1, Nodus 2, double precision Fehlberg method, relative error 10-8) :
VI-114 Nodus Examples Model V (mv) in compartment # 1 6 8/7/7 18/13/10 5a -19.30-19.01-18.95-18.90 5b -19.30-19.01-18.93-18.88 Error (%) +0.00 +0.00 +0.11 +0.12 5c -19.45-19.16-19.08-18.98 Error (%) +0.78 +0.80 +0.71 +0.45 The errors for current injections in the soma caused by the reduction are reasonably small in all compartments, including the soma. A -0.2 na, 400 ms current was injected in a branch (a: compartment #16,17 and 18/b: #11, 12 and 13/c:#10, Nodus 2, double precision Fehlberg method, relative error was 10-8). Model V (mv) in compartment # 1 6 8/7/7 18/13/10 5a -22.69-22.97-23.70-25.55 5b -22.65-22.94-23.86-26.12 Error (%) +0.14 +0.14-0.64-2.24 5c -22.78-23.07-23.99-29.87 Error (%) +0.42 +0.42 +1.21 +16.95 The errors for current injections in the branches caused by the reduction are reasonably small in the soma, but totally unacceptable at the reduced side-branches! Realistic Models Squid Giant Axon Demo This simulation data file contains the database for a simulation of the conduction of a classic Hodgkin and Huxley action potential ( HH Fast Na Currents and HH Delayed Rectifier ) in the Squid giant axon neuron model. An action potential current is triggered in compartment #1 by a 2 µa, 0.2 ms current pulse (see Current Clamp, the plot window (Fig. V/7) shows the membrane potential in several compartments (see Configure Plots). HH Fast Na Current and HH Delayed Rectifier These conductance definition files contain the equations developed by Hodgkin and Huxley for the fast inward sodium current and the delayed rectifier. The original equations (Hodgkin AL and Huxley AF: A quantitative description of membrane current and its application to conduction and excitation in nerve. J. Physiol. (London), 117 (1952) 500-544.) were adapted to the modern membrane potential convention (resting membrane potential is -70 mv). Squid Giant Axon Squid Giant Axon is a neuron definition file that contains a 10-compartment model of 10 mm of the axon that was modeled by Hodgkin and Huxley (Hodgkin AL and Huxley AF: A quantitative description of membrane current and its application to conduction and excitation in nerve. J. Physiol. (London), 117 (1952) 500-544.). All compartments have excitable membrane (the HH Fast Na Currents and HH Delayed Rectifier currents and a leak).
Nodus Examples VI-115 mv 40 20 0 5 10 15 20 ms -20-40 -60-80 Fig. VI/7: the plot output of the Squid Giant Axon Demo simulation. CS Fast Na Current, CS Delayed Rectifier and CS A Current These conductance definition files contain the equations for 3 voltage dependent ion currents. The equations are from De Schutter E: Alternative equations for the molluscan ion currents described by Connor and Stevens. Brain Res., 382 (1986) 134-138; and are derived from the paper Connor JA and Stevens CF: Prediction of repetitive firing behaviour from voltage clamp data on isolated neurone soma. J. Physiol. (London), 213 (1971) 31-53.
VI-116 Nodus Examples
Appendix VII-117 VII. APPENDIX Maxima for Memory Storage (Nodus 3.2) Simulations: Simulation database: Maximum number of simulation data files 1 Maximum number of simulation parameters in a database 20000 Maximum number of integrated variables in a database 4500 Maximum size of equation table 100000 Simulation plot: Maximum number of axes 4 Maximum number of plots on each axis 6 Maximum number of text output parameters 20 Experiments: Maximum number of current clamps 40 Maximum number of periods in a voltage clamp experiment 5 Maximum number of voltage clamped neurons 2 Maximum number of preset synaptic firing times 100 Maximum number of ionic conductance blocking periods 1 Networks: Maximum number of network definition files in memory 1 Maximum number of neurons in a network 200 Maximum number of connections from one presynaptic neuron 60 Neurons: Maximum number of neuron definition files in memory 20 Maximum number of compartments in memory: no 3-dimensional coordinates in use 4000 3-dimensional coordinates used in any neuron model 3000 Compartments: Maximum number of connections to other compartments: soma compartment 24 other compartments 6 Maximum number of ties to ionic current subdefinitions 1 Maximum number of ties to synaptic current subdefinitions 5 Maximum number of ties to transmitter release subdefinitions 1 Maximum number of ties to pool subdefinitions 2 Subdefinitions: Maximum number of ionic currents in memory 20 maximum number of currents in an ionic current subdefinition 13 Maximum number of synaptic currents in memory 60 Maximum number of transmitter release sites in memory 60 Maximum number of pools in memory 60 Conductances: Maximum number of conductance definition files in memory 20 Maximum number of (in)activation factors 2 The maxima are supplied as guidelines only. It is not guaranteed that these maximum numbers will remain unchanged in future updates of Nodus 3.
VII-118 Appendix Shift and Option Key Menu Modifications Only menu commands which can be modified are shown. Press the shift key and/or option key before selecting the menu command to get the required modification. No key Shift key Option key Shift+option keys File menu New Simulation New Simulation New Simulation New Simulation Open Open Open Open Close file name Close file name Hide file name Hide simulation Close file name Close simulation Kill file name Kill file name Close All Windows Hide All Windows Hide All Graphs Hide All Windows Neuron menu Go to Compartment Go to # Go to Compartment Go to # New simulation database uses the old database in memory, initial values are the old values. No dialog window is shown. Opens the definition file without creating a definition window. The data are loaded into memory and the file name is put at the bottom of the appropriate menu. The simulation database is closed, the plot window remains on the screen.
Appendix VII-119 Bugs and Problems Consult the ftp site (see Nodus 3.2 Update) for recent information about bugs and problems.
VII-120 Appendix Import Formats An import file is a standard Macintosh TEXT file, i.e. an ASCI file with only a carriage return (CR) at the end of a line (no linefeeds). Most entries are real or integer numbers and are separated by TABs and/or spaces. Genesis Genesis is a Unix neuron and network simulation program developed at the Claifornia Institute of Technology. It can read neuron morphology from.p files, which are partially supported by Nodus. The file consists of comment lines (preceded by // ), control lines (preceded by * ; only *symmetric, *absolute, *relative, *spherical, *cylindrical, *setglobal CM, *setglobal RM and *setglobal RA are supported) and compartment lines (one for each compartment). The compartment lines should be ordered by their connections (from parent to child). Nodus reads the 6 first items (separated by TABS or spaces) on compartment lines (channel objects are not supported) in the format: NAME PARENT X Y Z D. NAME is the name of the compartment. PARENT is the name of the parent-compartment to which this one is connected (by a node connection) or none. X, Y and Z are the center coordinates in µm of the distal face of the compartment cylinder and determine the compartment length (integer or real format). D is the diameter in µm (integer or real format). Ladder An extremely simple neuron morphology format. Each compartment is represented by one line of text, consisting of the diameter in µm and the length in µm, in integer or real format separated by TABS or spaces. Remark that connections between compartments cannot be specified, Nodus therefore assumes that they form a long, unbranched ladder. NINDS, NINDS with Rm The format used at the Laboratory of Neural Control, NINDS/NIH (Dr. Burke). The morphology is determined a branch and order number for each part of a dendrite. Branching is always binary and branch numbers depend on the number of the parent branch: BR child1 = 2*BR par -1 and BR chil d2 =2*BR par. This nomenclature is described in Segev I, Fleshman J W, Miller J P and Bunow B: Modeling the electrical behavior of anatomically complex neurons using a network analysis program: passive membrane. Biol. Cybern., 53 (1985) 27-40. There is one line of data for each segment of a branch, corresponding to one compartment in the neuron model. All entries are integer numbers, separated by TABS or spaces, with as format: ID ORD BR SEG TYPE X1 Y1 Z1 X2 Y2 Z2 LENGTH DIAMETER [RESISTANCE] ID: dendrite number (between 1 and 32000). ORD: branching order (between 1 and 50) BR: branch number (derived from parent branch, between 1 and 32000). SEG: sequential number when the branch is broken into subsegments (between 1 and 32000). TYPE: 1=next compartment is next segment of same branch. 2=branch point: next compartment is order+1.
Appendix VII-121 3=termination of branch. X1, Y1, Z1: 3 dimensional spatial coordinates of the start point of the segment. X2, Y2, Z2: 3 dimensional spatial coordinates of the end point of the segment. LENGTH: length of segment in µm*100. DIAMETER: diameter of segment in µm* 100. RESISTANCE: is optional, only in the NINCDS with Rm format; specific membrane resistance for this segment in kω*100. The file contains the data for all of the dendrites. The file does not have to be ordered, but conversion will go faster if it is. The first line always has to be the description of the soma, with ID=0, ORD=0, BR=0, SEG=0 and TYPE=0. The stem of a dendrite has ORD=0, BR=1 and SEG=1. Oxford The format used at the Laboratory of Neuropharmacology, University of Oxford (Dr. Somogyi). First line is the number of segment input lines (integer). Segment input lines describe very small parts of the (EM) measured neuron, several segments may be fused into one compartment in the neuron model. All entries are real numbers (except TYPE which is integer), separated by commas and spaces, with as format: X,Y,Z,TYPE,D X, Y, Z: coordinates of the segment in µm. TYPE: segment type code (integer value): 3: start. 4: continue. 5: fibre swelling, Nodus cosiders this to be the soma circumference. 6: spine base. 7: branch split. 8: branch end. D: diameter of the segment in µm. If no type code 5 is found in the import file, Nodus will take the first data point as the soma and use the diameter at this point. This may result in a soma which is too small. Also Oxford requires exact spacing between the data points (Fortran FORMAT type of formatting), an example of a correct line is: -0011.90,00049.80,-0020.00,005,3.50
VII-122 Appendix References A paper describing Nodus 2.3 has been published. Please refer to this paper when publishing results of modeling with Nodus. E. De Schutter: Computer software for development and simulation of compartmental models of neurons. Computers in Biology and Medicine 19: 71-81 (1989). Methodology of compartmental modeling Edwards, D. H., Jr., and B. Mulloney (1984) Compartmental models of electrotonic structure and synaptic integration in an identified neurone. J. Physiol. (London) 348: 89-113. (use of exponential peeling method to create 3-compartment models). Koch, C., and I. Segev (1989) Methods in neuronal modeling: from synapses to networks. MIT Press, Cambridge, MA. (very good book, overview of compartmental modeling and other realistic modeling methods with extensive examples). MacGregor, R. J. (1987) Neural and brain modeling. Academic Press, London. (mediocre book, overview of lots of modeling methods with extensive literature reviews). Perkel, D. H., and B. Mulloney (1978) Electrotonic properties of neurons: steady-stade compartmental model. J. Neurophysiol. 41: 621-639. (steady state analysis of compartmental models). Perkel, D. H., and B. Mulloney (1978) Calibrating compartmental models of neurons. Amer. J. Physiol. 235: 93-98. (steady state analysis of compartmental models). Perkel, D. H., B. Mulloney, and R. W. Budelli (1981) Quantitative methods for predicting neuronal behavior. Neuroscience 4: 823-837. (good introduction to passive compartmental modeling). Rall, W. (1962) Theory of physiological properties of dendrites. Ann. N.Y. Acad. Sci. 96: 1071-1092. (standard compartmental modeling reference!). Rall, W. (1964) Theoretical significance of dendritic trees for neuronal input-output relations. In Neuronal theory and modeling, Reiss, R. F., ed., pp. 73-97, Stanford University Press, Stanford. (standard compartmental modeling reference!). Shelton, D. P. (1985) Membrane resistivity estimated for the Purkinje neuron by means of a passive computer model. Neuroscience 14: 111-131. (extensive analysis of methodology and problems, comparison with equivalent cylinder model). Examples of compartmental modeling in invertebrate neurobiology Edwards, D. H., Jr., and B. Mulloney (1987) Synaptic integration in excitatory and inhibitory crayfish motoneurons. J. Neurophysiol. 57: 1425-1445. Getting, P. A. (1974) Modification of neuron properties by electrotonic synapses. I. Input resistance, time constant, and integration. J. Neurophysiol. 37: 846-857. Miller, J. P., and G. A. Jacobs (1984) Relationships between neuronal structure and function. J. Exp. Biol. 112: 129-145. Waldrop, B., and R. M. Glantz (1985) Synaptic mechanisms of a tonic EPSP in crustacean visual interneurons: analysis and simulation. J. Neurophysiol. 54: 636-650. Zucker, R. S. (1972) Crayfish escape behavior and central synapses. III. Electrical junctions and dendrite spikes in fast flexor motoneurons. J. Neurophysiol. 35: 638-651. Examples of compartmental modeling in vertebrate neurobiology Clements, J. D., and S. J. Redman (1989) Cable properties of cat spinal motoneurones measured by combining voltage clamp, current clamp and intracellular staining. J. Physiol. (London) 409: 63-87.
Appendix VII-123 Fleshman, J. W., I. Segev, and R. E. Burke (1988) Electrotonic architecture of type-identified alpha-motoneurons in the cat spinal cord. J. Neurophysiol. 60: 60-85. Gamble, E., and C. Koch (1987) The dynamics of free calcium in dendritic spines in response to repetitive synaptic input. Science 236: 1311-1315. Holmes, W. R., and C. D. Woody (1989) Effects of uniform and non-uniform synaptic activation-distributions on the cable properties of modeled cortical pyramidal neurons. Brain Res. 505: 12-22. Knowles, W. D., R. D. Traub, R. K. S. Wong, and R. Miles (1985) Properties of neuronal networks: experimentation and modeling of the epileptic hippocampal slice. Trends Neurosci. 8: 73-79. Koch, C. (1985) Understanding the intrinsic circuitry of the cat's lateral geniculate nucleus: electrical properties of the spine-triad arrangement. Proc. Roy. Soc. London Ser. B 225: 365-390. Llinas, R., and C. Nicholson (1976) Reversal properties of climbing fiber potential in cat Purkinje cells: an example of a distributed synapse. J. Neurophysiol. 39: 311-323. Miller, J. P., W. Rall, and J. Rinzel (1985) Synaptic amplification by active membrane in dendritic spines. Brain Res. 325: 325-330. Nitzan, R., I. Segev, and Y. Yarom (1990) Voltage behavior along the irregular dendritic structure of morphologically and physiologically characterized vagal motoneurons in the guinea pig. J. Neurophysiol. 63: 333-346. Pellionisz, A., and R. Llinas (1977) A computer model of cerebellar Purkinje cells. Neuroscience 2: 37-48. Perkel, D. H. (1983) Functional role of dendritic spines. J. Physiol. (Paris) 78: 695-699. Perkel, D. H., and D. J. Perkel (1985) Dendritic spines: role of active membrane in modulating synaptic efficiency. Brain Res. 325: 331-335. Pun, R. Y. K., E. A. Neale, P. B. Guthrie, and P. G. Nelson (1986) Active and inactive central synapses in cell culture. J. Neurophysiol. 56: 1242-1256. Rall, W., and G. M. Shepherd (1968) Theoretical reconstruction of field potentials and dendrodendritic synaptic interactions in olfactory bulb. J. Neurophysiol. 31: 884-915. Schierwagen, A. (1986) Segmental cable modelling of electrotonic transfer properties of deep superior colliculus neurons in the cat. J. Hirnforsch. 27: 679-690. Segev, I., J. W. Fleshman, J. P. Miller, and B. Bunow (1985) Modeling the electrical behavior of anatomically complex neurons using a network analysis program: passive membrane. Biol. Cybern. 53: 27-40. Segev, I., and W. Rall (1988) Computational study of an excitable dendritic spine. J. Neurophysiol. 60: 499-523. Shepherd, G. M., R. K. Brayton, J. P. Miller, I. Segev, J. Rinzel, and W. Ral (1985) Signal enhancement in distal cortical dendrites by means of interactions between active dendritic spines. Proc. Natl. Acad. Sci. USA 82: 2192-2195. Stratford, K., A. Mason, A. Larkman, G. Major, and J. J. B. Jack (1989) The modelling of pyramidal neurones in the visual neurones in the visual cortex. In The computing neuron, Dubin, R., C. Miall, and G. Mitchison, eds., pp. 296-321, Addison-Wesley, Wokingham, UK. Traub, R. D. (1982) Simulation of intrinsic bursting in CA3 hippocampal neurons. Neuroscience 7: 1233-1242. Traub, R. D., F. E. Dudek, R. W. Snow, and W. D. Knowles (1985) Computer simulations indicate that electrical field effects contribute to the shape of the epileptiform field potential. Neuroscience 15: 947-958.
VII-124 Appendix Traub, R. D., F. E. Dudek, C. P. Taylor, and W. D. Knowles (1984) Simulation of in vitro synchronized hippocampal discharges occuring in the absence of chemical synaptic transmission. Abstr. Soc. Neurosci. 10: 548-548. Traub, R. D., F. E. Dudek, C. P. Taylor, and W. D. Knowles (1985) Simulation of hippocampal afterdischarges synchronized by electrical interactions. Neuroscience 14: 1033-1038. Traub, R. D., W. D. Knowles, R. Miles, and R. K. S. Wong (1984) Synchronized afterdischarges in the hippocampus: simulation studies of the cellular mechanism. Neuroscience 12: 1191-1200. Traub, R. D., and R. Llinas (1979) Hippocampal pyramidal cells: significance of dendritic ionic conduc tances for neuronal function and epileptogenesis. J. Neurophysiol. 42: 476-496. Traub, R. D., R. Miles, R. K. S. Wong, L. S. Schulman, and J. H. Schneiderman (1987) Models of synchronized hippocampal bursts in the presence of inhibition. II. Ongoing spontaneous population events. J. Neurophysiol. 58: 752-764. Traub, R. D., R. Miles, and R. K. S. Wong (1987) Models of synchronized hippocampal bursts in the presence of inhibition. I. Single population events. J. Neurophysiol. 58: 739-751. Traub, R. D., R. Miles, and R. K. S. Wong (1989) Model of the origin of rhythmic population oscillations in the hippocampal slice. Science 243: 1319-1325. Traub, R. D., and R. K. S. Wong (1983) Synaptic mechanisms underlying interictal spike initiation in a hippocampal network. Neurology 33: 257-266. Traub, R. D., and R. K. S. Wong (1983) Synchronized burst discharge in disinhibited hippocampal slice. II. Model of cellular mechanism. J. Neurophysiol. 49: 459-471. Traub, R. D., R. K. S. Wong, R. Miles, and W. D. Knowles (1985) Neuronal interactions during epileptic events in vitro. Fed. Proc. 44: 2953-2955. Other modeling software Bunow, B., I. Segev, and J. W. Fleshman (1985) Modeling the electrical behavior of anatomically complex neurons using a network analysis program: excitable membrane. Biol. Cybern. 53: 41-56. Carnevale, N. T., and F. J. Lebeda (1987) Numerical analysis of electrotonus in multicompartmental neuron models. J. Neurosci. Meth. 19: 69-87. Hines, M. (1989) A program for simulation of nerve equations with branching geometries. Int. J. Biomed. Comput. 24: 55-68. Wilson, M. A., U. S. Bhalla, J. D. Uhley, and J. M. Bower (1989) GENESIS: a system for simulating neural networks. In Advances in neural information processing systems, Touretzky, D., ed., pp. 485-492, Morgan Kaufmann, San Mateo, CA.
Appendix VII-125
VII-126 Appendix NODUS 3.2 UPDATE User Interface The most obvious change is the new style user interface. All dialog windows and popup menus now use the Geneva 10 font instead of Chicago 12. This change was implemented after a survey showed that the majority of Nodus 3.1 users would prefer a smaller font size. The advantage is that more text can be shown for the same window size. It also looks cool. However, if you don't like the small fonts, there is an option in the Preferences command (Edit menu) which allows you to switch to the old-style System 6 window display. You will have to restart Nodus before this takes effect. If you use the old-style display, you will notice that some dialogs have become quite crowded and that some new options have not been implemented in the old style dialogs. Also, if you still use System 6, the old-style dialogs will be shown. The preferences file (Nodus 3.1 manual, p. 29) has been renamed to Nodus_Preferences. This file works only with Nodus 3.2, if you want to continue using Nodus 3.1 you should keep the it's preference file ('Nodus Preferences'). Not only does this new name distinguish if from the old preferences file, but it also allows me to e-mail the preferences file to new users. The Nodus_Preferences' file should be in the same folder as Nodus 3.2 or in the Preferences folder inside the System Folder. It is a personalized file, so you should not give it to other people! Scrolling of graphs has been improved. Copy, Cut and Paste (Edit menu) are now available for all text items in dialog and definition windows. Copy of PICT data from a graphics window remains functional (Nodus 3.1 manual, p. 62). Balloon Help (see your Macintosh manual) has been implemented. All menu commands have balloons. At present only a few dialog windows have balloons implemented, but in the near future all dialog windows will have balloons. Balloon Help needs the Nodus Help file, which should be in the same folder as Nodus. Otherwise, the Help menu will show an Open Nodus Help command which allows you to open the Nodus Help file anyway. File System Nodus 3.2 can read all Nodus 3.1 files (and Nodus 2 neuron and conductance definition files). Nodus 3.1 simulation data files can be read, but no simulation window will be created; you have to do New Simulation and recompile the simulation database (Nodus 3.1 manual, p.41). Nodus 3.2 can save to a new file format. It can save any graph window to a PICT file. PICT files are a standard format for graphical data on Macintosh computers and can be Opened by most drawing, painting and page setting applications. If the active window is a conductance or neuron plot or an old simulation window, the Save As command will change into Save As PICT File. If you want to save the active simulation plot window as a PICT file, you have to press the option key while selecting the File menu. Nodus 3.2.1 cannot Open PICT files. Another change is the use of smart links, which uses advanced features of the System 7 filing system. As a consequence, linked files (see page 30 of the Nodus 3.1 manual) can now be put anywhere on your hard disk. There is no longer any need to keep linked files together in the same folder so that they can be Opened together automatically. The Nodus Data folder has been reorganized; all conductance files have been put together in a folder, as have been all neuron and network files. The System 7 filing system is quite smart. You can move files around to other folders or even rename them and Nodus will still find the linked files automatically, provided that file names are unique. This is a very important caveat: never have two Nodus files with the same name on your disk! Also, never Duplicate (or drag and copy) your Nodus files, except for making a backup copy to another disk, because duplicated files will have the same hidden file-id number (see page 30 of the Nodus 3.1 manual) so that Nodus cannot distinguish them. Note also that the Save As command is not the proper way to move a file from one folder to another, instead
Appendix VII-127 you should Quit Nodus, go to the Finder and drag the file to it's new folder (if use Save As the file is given a new hidden file-id number and considered a new file by Nodus). If you want to have multiple copies of the same file in different folders (for example a conductance definition), you can use the Make Alias command in the Finder (File menu) and put aliases of the file in different folders. Like other applications, Nodus does not distinguish aliases from the original file. Finally, note that this 'smart linking' only works for files that have been saved by Nodus 3.2, Nodus 3.1 files behave as before. Because of smart linking, you can put Nodus files in separate folders. I suggest that you do not store your own Nodus files in the Nodus Data folder, but use your own folder. This will make it easy to update the Nodus Data folder when new example files become available: you can then drag your old copy of the Nodus Data folder into the Trash and install the new 'Nodus Data' folder without loosing any of your own files. If you still use System 6, you should reorganize the Nodus Data folder. drag all files out of the subfolders (Conductances, etc.) and put them together in one folder. Conductance Definitions Nodus 3.2 accepts a much wider choice of conductance equations than Nodus 3.1. First of all, conductances can now depend on voltage, on ionic concentration, or on both. Second, instead of using a Hodgkin Huxley-type equation, you can tabulate the conductance variables in a text file. These additions resulted in a major redesign of the conductance definition window (see also the Nodus 3.1 manual, page 99). In system 7 the definition window shows both the activation and inactivation equations (if there is no inactivation, only activation is shown), each in it's own box. At the top left of each box a popup menu allows you to select the type of conductance equation. Basically three types are supported at present: standard Hodgkin Huxley equation, a table file containing alpha and beta and a table file containing steady state (in)activation (Minf) and time constant (tau, see Nodus 3.1 manual, p. 20). Each of these types can be either voltage-dependent, concentration- dependent, or both. If you select the table file option, you have to prepare the table outside of Nodus. A spreadsheet application, like Microsoft Excel or Lotus 1,2,3, makes it easy to create these files if you use equations, just save it as a text file. Of course, you can also use experimental data for Minf and tau and use a word processing application to create the table file. In the conductance definition file a button appears, which allows you to select the table file with the standard Open file dialog. Table files are read by Nodus when it creates the equation table for a New Simulation (Nodus 3.1 manual, p. 69). Because Nodus will usually store the conductance variables for different voltages values than the ones listed in the table, interpolation is necessary. As default, Nodus uses cubic spline interpolation, but this can be quite slow, especially for large two-dimensional tables. Therefore, the Preferences command (Edit menu) has a Fast interpolation of tables option. When fast linear interpolation is selected, the equation table will be computed much more rapidly, but less accurately. Note also that the maximum errors shown for during creation of the equation table apply only to standard Hodgkin Huxley equations, not to table files (because they are usually not smooth functions). For voltage- or conductance- dependent equations, the table file should contain 3 columns: the first column is voltage or conductance, the second alpha or Minf and the third beta or tau. Columns can be separated by tabs or spaces. They must be in increasing order of voltage (i.e. from hyperpolarized to depolarized) or conductance, but the increase between rows must not be constant. This allows you to specify outlying values or to use smaller steps when the relation between variable and voltage or conductance is very steep. The first row contains a header consisting of 2 integer values, i.e. the number of rows (excluding the header), followed by the number of dependent variables (usually 2).
VII-128 Appendix For mixed voltage- and conductance dependent equations, the table file is twodimensional, with each row a different voltage and each column a different concentration. Usually two tables are put next to each other, first come all the columns for alpha or Minf, then all columns for beta or tau. The header consists of two rows. The first row of the header contains 3 integer values, i.e. the number of rows (excluding the header), the number of concentration values (this is usually the number of columns minus one, divided by two) and the number of dependent variables (usually 2). The second row of the header contains the concentration values. Note that two-dimensional tables use a lot of memory and may take a lot of time to create. Therefore, Nodus imposes a maximum of 100 columns on these tables, so you should not use more than 100 concentration values in the table file. Several examples of table files are supplied in the Nodus Data folder. See the Traub82 and Yamada series of conductance definition files. Finally, you can use two options in the conductance definition window to change the use of time constants for some of the conductance equation types. Conductances determined by steady state (in)activation and time constant, you can be made instantaneous. This means that there is no time constant (if you use a table file, you have only 1 dependent variable, i.e. Minf) and that (in)activation always equals steady state (in)activation. You can also specify that the (in)activation should have a second time constant. This second time constant (tau2) is not completely independent of the first time constant (tau1, computed by the conductance equation), as tau2 = tau1 * factor, where factor equals the tau2/tau1 which you specify in the conductance definition window. You also need to specify the fraction of (in)activation gates that uses the second time constant, activation will be computed as: M = (1- fraction)*m1 + fraction*m2, where M1 is activation using tau1 and M2 is activation using tau2. Concentration Pools To use concentration-dependent conductances, you also must compute concentrations. This has been implemented in Nodus 3.2 as the pool subdefinition, which uses a wide range of modeling features, including diffusion, buffers and Nernst potentials. For more information and a description of the equations involved see the chapter by Yamada et al. in Methods in Neuronal Modeling by Koch and Segev (1989). Pools are defined as any other subdefinition with the new Pools command (Neuron menu). Like other subdefinitions, they must be tied to a compartment to become functional, see Nodus 3.1 manual p. 33-35 for a description of subdefinitions. However, pool subdefinitions can be linked to each other into groups, which makes them more complex than other subdefinitions. Imagine an onion shell type of model for diffusion, with ten shells. You wouldn t want to tie all ten shells individually to each compartment. Therefore you can group these shells together, by linking them, and tie the first one of the group to the compartment. The pool subdefinition dialog window has at the top left a Type popup menu. Pool subdefinitions are either a pool, which is a volume of unspecified shape, or a shell. For shells you have to specify the position relative to the membrane of the compartment. Each group can contain one supramembrane shell (i.e. the volume just outside of the membrane) and one submembrane shell (i.e. the volume just inside the membrane), you need both if you want to use Nernst potentials. If you want to model diffusion, you can define additional shells, 'inside shells' if you want to model diffusion into the cell, outside shells if you want diffusion away from the cell. You can use as many inside or outside shells as you like, up to a maximum of 16 shells in one group. If you use 'inside shells, you probably also want to use the core which is the remaining volume in the center of the compartment.
Appendix VII-129 Pool subdefinitions can be linked together in groups. Use the Link to popup menu (top right) to link a child pool subdefinition to a parent (note that this is relation is opposite to the way you link compartments in the compartment definition). Each child can have only one parent and vice versa. They are all linked together in a group, named for the first parent (as shown in the Group name at the top right of the dialog window), which can be tied to a compartment. Diffusion is possible only to a parent and to a child. Each pool has a Size (the thickness of the shell or the volume of the pool) and a Minimum concentration (which is also the initial concentration in µm). Each pool can have several Actions. If no actions are defined, concentration is constant (this can be useful for a Nernst potential). You select an action with a popup menu, up to 10 different actions can be defined for each pool subdefinition. Most action cause the concentration in the pool to change, except for conductance (in)activation. Each action type needs 1 to 3 parameters. Because different types of actions are all listed together, it was impossible to show proper titles for all the action parameters. The titles shown reflect the last action type that was selected. To show the proper parameter titles for any action, press it s popup menu or any of it's parameters. The decay by tau action causes concentration to decay exponentially to the minimum concentration. The decay rate is determined by a time constant (in ms). This simple method of modeling changes in concentration is often used when detailed simulation of calcium is not deemed necessary. See for example the Traub82 Hippocampal Neuron. However, it can also be used to implement a simple model of an ion pump. Only one decay by tau action can be specified for a pool subdefinition. The diffusion action causes diffusion to a neighboring shell or pool. Diffusion is only possible to linked pool subdefinitions (i.e. the parent or child ) and is always bidirectional. You have to specify the pool subdefinition to diffuse to, and the diffusion constant (in µm 2 /ms). For pools a coupling factor must also be supplied, for shells Nodus computes the coupling factor based on the size of the compartment and the shell thickness. The ionic current flow action causes the flow through the specified ionic channel to change the concentration. You have to specify the channel (with a popup menu). Parameters are the fraction of the current that contributes to the concentration (default is 1) and the ionic valency (e.g. +2 for calcium, +1 for potassium). The nernst + ionic current action affects the concentration pool in the same way as the ionic current flow action, but the ionic current is computed differently. It's reversal potential is no longer constant (as specified with the Ionic Currents command, Nodus 3.1 manual, p. 92), but determined by the Nernst potential. To compute the Nernst potential, both a supramembrane and a submembrane shell must be tied to the compartment (they may belong to the same group of pool subdefinitions or not), and at least one of them must have the nernst + ionic current action. The GHK + ionic current action affects the concentration pool in the same way as the ionic current flow action, but the ionic current is computed differently. Ionic current is no longer a linear function of voltage, but is instead computed by the Goldman-Hodgkin-Katz equation. To use the GHK equation, both a supramembrane and a submembrane shell must be tied to the compartment (they may belong to the same group of pool subdefinitions or not), and at least one of them must have the GHK + ionic current action. Not implemented in Nodus 3.2.1. The synaptic current flow action causes the flow through the specified synaptic channel to change the concentration. You have to specify the channel (with a popup menu). Parameters are the fraction of the current that contributes to the concentration (default is 1) and the ionic valency (e.g. +2 for calcium, +1 for potassium).
VII-130 Appendix The nernst + synaptic current action affects the concentration pool in the same way as the synaptic current flow action, but the synaptic current is computed differently. It s reversal potential is no longer constant (as specified with the Synaptic Currents command, Nodus 3.1 manual, p. 96), but determined by the Nernst potential. To compute the Nernst potential, both a supramembrane and a submembrane shell must be tied to the compartment (they may belong to the same group of pool subdefinitions or not), and at least one of them must have the nernst + synaptic current action. The GHK + synaptic current action affects the concentration pool in the same way as the synaptic current flow action, but the synaptic current is computed differently. Synaptic current is no longer a linear function of voltage, but is instead computed by the Goldman-Hodgkin-Katz equation. To use the GHK equation, both a supramembrane and a submembrane shell must be tied to the compartment (they may belong to the same group of pool subdefinitions or not), and at least one of them must have the GHK + synaptic current action. Not implemented in Nodus 3.2.1. The buffer #1, buffer #2 and buffer #3 actions are identical. They have been named this way to allow output of the free buffer concentrations. In the Configure Plots ( Nodus 3.1 manual, p. 70) and Text Output (Nodus 3.1 manual, p. 72) commands, the Value popup menu includes options to select free buffer #1, etc. These actions implement first order buffering of the concentration in the active pool subdefinition. Parameters are the total buffer concentration (in µm), the forward rate of buffer binding (in (1/µM ms) and the backward rate of buffer binding (in 1/ms). Each of these actions can be used only once in a pool subdefinition. The pump action implements an ion pump. Three parameters are necessary: a Kmax (in 1/ms), a Kd (in µm) and a density (in µmol/µm 2 ). See Zador et al., PNAS 87, 6718-6722, 1990 for the equation. The conductance (in)activation action is shown in italic to emphasize that it is different. It means that the active pool subdefinition will be used for the concentrationdependent (in)activation of the selected conductance. The conductance (in)activation action is available only in pools and submembrane shells. You have to specify the conductance equation with a popup menu (all voltage-dependent conductances are disabled) and the maximum range for the equation table (see Nodus 3.1 manual, p. 20, the table will contain conductance variables in the range 0 µm to the maximum you specified for the first conductance (in)activation action encountered during simulation database compilation). Subdefinition management: is similar to that for other subdefinitions. Additionally, there is a Smart duplicate & link button. This button is useful for duplicating shells if you want to create an onion model. The duplicated shell will be linked to the original one and have the same actions (if appropriate). Neuron Definitions Apart from pool subdefinitions (see above), other changes are: Current subdefinitions can have up to 13 currents defined. You can have now up to 5 different ties to synaptic current subdefinitions in the compartment definition window (Nodus 3.1 manual, p. 87). This, together with the new Strength field for Synaptic Connections (see next) expands the network capacity considerably. You can also have up to 2 ties to pool subdefinitions. A Variable option has been added to Membrane capacitance in the neuron definition window (Nodus 3.1 manual, p.85). This option is similar to the Variable option for Membrane resistance, e.g. you set the value in the compartment definition window. Note that either capacitance or resistance can be variable, but not both at the same time. However, one can combine variable membrane capacitance with 2 values for membrane resistance (the From compartment # option).
Appendix VII-131 Network Definitions Nodus 3.2 can accommodate much larger networks than before: up to 200 neurons, with up to 60 connections from each neuron. A field has been added to the Synaptic Connections (Network menu, Nodus 3.1 manual, p. 82) dialog. The Strength field allows you to give specific synaptic connections different strengths, default is one. The peak synaptic conductance Gpeak is now: Gpeak = Gmax * Strength * Transmitter_amount Future Nodus releases will implement synaptic plasticity by dynamic changes in the Strength field, Note that one can already model short term potentiation by using a presynaptic calcium pool that controls the amount of transmitter released. Nodus 3.2 implements gap junctions, which can be both rectifying and non-rectifying. A new command Electrical Connections in the Network menu allows you to set up these gap junctions. The number of electrical connections from each neuron is also listed in the network dialog window. The approach is very similar to the one used in the Synaptic Connections command, i.e. you select a neuron from which the electrical connection will start with the from popup menu at the top. Then you can specify up to 20 connections by selecting a to neuron with a popup menu. The design of the Electrical Connections dialog differs a bit because a compartment preselection popup menu (Nodus 3.1 manual, p. 48) is included for both the from and o neuron. This reflects the fact that these connections go from one compartment to another. You also have to set the conductance G of the connection (in ns). This implements a symmetric, non-rectifying gap junction. To make it rectifying, you first have to define a conductance (New Conductance) that describes the voltage (or concentration) dependence of the rectification. You can then select this conductance from the rectification popup menu to make the electrical connection rectifying. The current flowing through the gap junction is computed as: Ig = Gg * (Vpost - Vpre) Gg is constant (non-rectifying) or a function of Vpre (rectifying). Nodus 3.2.2 can randomize synaptic weights with the Randomize Strenghts command in the Network menu. Randomize synaptic connection weigths and/or electric connection conductances in a network. Randomization can be either as a uniform distribution (specify the range) or as a gauss distribution (specify mean and standard deviation). The new weights will depend on the old ones if the Multiply with existing values option is selected. Integration The integration routines have changed quite a lot internally compared to Nodus 3.1. As a consequence, the same simulation may run a bit faster (especially large neurons) or slower (active membrane models in Euler method) and you can expect small differences in the results. In the Euler method (used to be called 'Hybrid Euler'), the changes in variable time step are controlled by a different mechanism. This results in slightly different minor time steps (Nodus 3.1 manual, p. 25), which may cause different results when large time steps are used. Also, if you use a model with pool subdefinitions, you will have to specify both a maximum change in voltage and in concentration per minor time step (Nodus 3.1 manual, p. 69) if you use the Euler method. In the Fehlberg method, conductances are now also interpolated from the equation table (Nodus 3.1 manual, p. 20), instead of computed by the Hodgkin Huxley equation for each time step. Usually this will cause no differences, because linear interpolation is correct if the table uses small voltage steps, but in some active membrane models you might note small differences.
VII-132 Appendix The voltage range for the equation table (Nodus 3.1 manual, p. 69) are no longer specific to each simulation database. Consequently, this range is not shown in the Integration Settings dialog (Simulation menu) anymore. You can change the voltage range in the Preferences command (Edit menu), such a change will apply to all simulations. Overview Of Changes In Menu Commands File Menu New Simulation The New Simulation command is now always available if a simulation database or neuron definition file is in memory. In Nodus 3.1 you had to Close an active simulation first before you could do New Simulation. This is no longer true, if a simulation database is still active it will be automatically closed by the New Simulation command before it displays it s dialog window. One can now Seed random number generator in the New Simulation dialog window. Open Simulation The Open Simulation command is now always available In Nodus 3.1 you had to Close an active simulation first before you could do Open Simulation. This is no longer true, if a simulation database is still active it will be automatically closed by the New Simulation command before it displays it s dialog window. Import Neuron Is no longer a hierarchical menu (Import Simulation has been removed). Two new import formats: NINDS (no coordinates) (is similar to NINDS (with coordinates), see Nodus 3.1 manual, p. 118, but without the X1, X2, X3, Y1, Y2, Y3 fields) and Eutectics (the popular reconstruction software, note that this import filter is still experimental, if it fails let me know). Save As a neuron definition file The Text file option was available in the latest Nodus 3.1 versions, but is not described in the Nodus 3.1 manual. One can save the neuron definition file to a text file with all the connections between compartments listed (Text with links, this is similar to the printed output) or without listing the connections (Text without links). I both cases all compartment data are separated by TABs, so the file can be easily imported into a spreadsheet or statistics application. For text files Nodus uses the following numbers to describe the type of connection (link) between compartments: 1: a node connection (parent to child) 2: a branch connection (parent to child) 3: a node connection (child to parent) 4: a branch connection (child to parent) Save As PICT file Variant of Save As command. See section on File System. Revert: New command. Deletes all changes you made to a definition and reloads it from the file. This command is available only after you have made changes to the active definition and if those changes have been stored in memory. Print a neuron definition file The complete printing of neuron definition files was available in the latest Nodus 3.1 versions, but is not described in the Nodus 3.1 manual.
Appendix VII-133 The complete neuron definition file contents are printed. This includes the cable parameters and comment; the name, shape, size, electrotonic length, local membrane capacitance or resistance, coordinates (if present), ties to subdefinitions (numbered) and links to other compartments of all compartments; and all subdefinitions. While printing neuron definition files Nodus uses the following symbols to describe the type of connection (link) between compartments: -o- a node connection (parent to child or vice versa) -> a branch connection (parent to child) <- a branch connection (child to parent) Edit Menu Copy The Copy command works for text, see section on User Interface. Cut The Cut command works for text, see section on User Interface. Paste The Paste command works for text, see section on User Interface. Duplicate When active, this command will show either Duplicate Neuron, Duplicate Compartment or Duplicate Conductance, depending on what type of definition window is active. Duplicate Neuron makes a copy of the active neuron definition in memory and shows its definition window, it is very similar to doing a Save As of a neuron definition (Nodus 3.1 manual, p. 30-31), except that no new file is created till you Save the copy. Duplicate Compartment is a much more useful command. It makes a copy of the active compartment and inserts it behind the active compartment (or appends it). This is very useful when you are creating a neuron model that has a lot of similar sized compartments, as the new compartment will have the same size and attributes as the active compartment. Branches linked to the active compartment are not copied. If the next compartment (by number) had not been defined yet, than Duplicate Compartment overwrites that compartment, otherwise the new compartment is inserted and the number of compartments increased by one. Duplicate Conductance makes a copy of the active conductance definition in memory and shows its definition window, it is very similar to doing a Save As of a conductance definition (Nodus 3.1 manual, p. 30-31), except that no new file is created till you Save the copy. Preferences Several new options have been added to the Preferences command. System 6 (large fonts): see section on User Interface. Fast interpolation of table: see section on Conductance Definitions. Equ. table from: see section on Integration. Simulation Menu Start/Pause/Continue The Run command (Nodus 3.1 manual, p. 68) has been renamed. The command is now called Start if the simulation has not been started yet and Continue after a Pause. The functionality has not been changed. Pause At New command. Let s you specify a time at which the simulation will pause. You can then change experiment or Text Output settings and Continue. Integration method The 'Equation table from' has been moved to the Preferences command, see section on Integration.
VII-134 Appendix Configure Plots: A Legend option has been added. It prints the variable type (in black) and names (in color) in the Simulation Plot window. You can now specify up to 6 plots for each axis. Text Output: The 'Make legend' option has been renamed Legend. It outputs 4 rows names, corresponding to the names of the value type (first row), followed by the neuron name, compartment name and subdefinition name. Legend is enabled only when Separated by TABs option is active. You can now specify up to 6 values for each value type. Current Clamp: This command is disabled if a Voltage Clamp is active in a single neuron simulation (and vice versa). 'Cyclical' has been renamed Repetitive, functionality has not changed. White noise currents (Nodus 3.1 manual, p. 79) have now a gaussian distribution, with mean zero. Instead of an Amplitude you have to specify the standard deviation Stand. Dev. and the 'Seed' is no longer specified, but you can set the seed for the random number generator in the New Simulation command. File input for Current Clamp. A disk icon has been added to the Current Clamp command. It allows you to specify a file from which current injection values will be taken. You also can specify an index into the file (if it contains more than 2 columns). The file should have the format: time0 current1.0 current2.0... time1 current1.1 current2.1...... timen current1.n current2.n ^ ^ index 1 index 2 Columns can be separated by spaces or tabs. You can use any time interval or starting time. Current will be zero before time0 and equal current#.n after timen (as long time is before the End specified in the Current Clamp dialog). No interpolation is performed. Voltage Clamp This command is disabled if a Current Clamp is active in a single neuron simulation (and vice versa). Synaptic Firing Times: A Random firing at option has been added. Click on this option and a edit field will appear in which you can specify the mean firing frequency (in Hz) for the selected synapse. Synapse will fire using a Poisson distribution with the specified mean. The random number generator can be seeded in the New Simulation command. Network Menu Synaptic Connections A Strength field has been added, see section on Network Definitions. The Delete all from this neuron button deletes all synaptic connections for which the current from neuron is the presynaptic neuron. Electrical Connections New command. See section on Network Definitions.
Appendix VII-135 Randomize Strenghts New command. See section on Network Definitions. Neuron Menu Go to Compartment The dialog has now also preselection and compartment popup menus (Nodus 3.1 manual, p. 28) that allow you to select the compartment. These popup menus become inactive if you type in a number. Sholl Plot This command was available in the latest Nodus 3.1 versions, but is not described in the Nodus 3.1 manual. Sholl Plot draws a neuron diagram of the active neuron definition file in a graphics window with scroll bars. The command is available only for neuron definitions with the Tree model format option active. One can Copy, Save As PICT or Print the neuron diagram. The layout of the Sholl plot is under user control. The dialog window has the following options: Layout: either a Horizontal (soma at left) or Vertical (soma at bottom) Sholl plot is drawn. Lengths: the morphological, Real length (in µm) of the compartments is shown or their Electrotonic length. You also have the option of loading values From a text file (a single column, one value for each compartment). Diameters: are Not shown (compartments are drawn as line segments, this is the standard Sholl plot) or the morphological, Real diameters are shown (compartments are drawn as rectangles). Spherical compartments are always shown as circles. Expand weight factors: if enabled any branch connected with a weight factor of n, larger than one, will be drawn n times. If not enabled all compartments are drawn only once. Branch connections are always drawn differently from node connections: the branch compartment is connected to the center of the parent compartment. Scale: the size of the Sholl plot is determined by this setting. For Real Lengths the units are pixels/µm (a pixel is 1 point on the screen, most Mac screens have a resolution of 72 pixels per inch), for Electrotonic Lengths the units are pixels/millilamba (i.e. per 0.001 lamba). Because of minimum spacing between branches requirements, the Sholl plot will always have a minimum vertical (Horizontal Layout) size. If the Sholl plot is much larger than your screen, it might help to Copy it and Paste it in a graphics application window (like MacDraw) that allows you to zoom out. Transmitter Release Transmitter release can now be Concentration pool dependent. You also have to specify a scaling Factor and a Power (e.g. it is often assumed that transmitter release is proportional to the third power of the calcium concentration). Synaptic Currents The implementation of conductance dependent synapses has been modified. Conductance dependence is now an extra property that a synapse can have. The time course of the synaptic current is determined by the Alpha or Dual exponential function, but if Also conductance dependent is selected, it's peak conductance will be determined by the conductance equation. An example of a NMDA synapse is included (neuron definition file Spine with NMDA channel). Pools New command. See section on Concentration Pools Conductance Menu Conductance Plot windows are now recognized as belonging to a conductance definition and activate the Conductance menus.
VII-136 Appendix Plot Scales 'Potential range' has been renamed Variable range and a 0 to... µm field has been added. This controls the range for concentration dependent conductances. For conductances that are both potential and concentration dependent, both variables change together; you probably want to constrain one variable by giving it a very narrow window. The Overlay existing plots button is now active. Note that this option must have been selected before you draw the original plot.
Appendix VII-137 Nodus ftp site Instructions for ftp: Most of the files at the ftp site have been compressed and binhexed with StuffIt 1.5. The files are marked by a double suffix: '.sit' for the compression and '.hqx' for the binhex conversion. Some ftp software (like Fetch) automatically decodes and decompresses the files after downloading them, I strongly advise you to use such software. Otherwise, download the files, and use either StuffIt 1.5, StuffIt Lite, StuffIt Deluxe or another compatible application to decompress these files. First Decode BinHex File the file (this strips the '.hqx' suffix) and then Open Archive the file, select the content and press the Extract button. You can also consult our www page at http://bbf-www.uia.ac.be/ Using the ftp site The Nodus server is a unix machine called 'bbf-ftp.uia.ac.be' (143.169.8.193). You can access it by standard anonymous ftp. Detailed instructions for ftp from another unix machine: yourhost% "ftp bbf-ftp.uia.ac.be" Connected to 143.169.8.193. 220 kuifje FTP server (SunOS 4.1) ready. Name (143.169.8.193:yourname): "anonymous" 331 Guest login ok, send ident as password. Password: "yourname@yourhost.yourside.yourdomain" 230 Guest login ok, access restrictions apply. ftp> "cd nodus" 250 CWD command successful. ftp> "ls" 200 PORT command successful. 150 ASCII data connection for /bin/ls (131.215.137.243,1682) (0 bytes). NODUSINFOTEXT Nodus3.2.2.sit.hqx Nodus3.2.2Q.sit.hqx NodusData.sit.hqx NodusHelp.sit.hqx NodusInfoWord.sit.hqx README demo_version_only 226 ASCII Transfer complete. 144 bytes received in 0.02 seconds (6.1 Kbytes/s) ftp> "binary" 200 Type set to I. ftp> "mget Nodus*" mget Nodus3.2.1.sit.hqx? "y" 200 PORT command successful. 150 Binary data connection for Nodus3.2.1.sit.hqx (131.215.137.243,1690) (712750 bytes). 226 Binary Transfer complete. local: Nodus3.2.2.sit.hqx remote: Nodus3.2.1.sit.hqx 712750 bytes received in 5.8e+02 seconds (1.2 Kbytes/s) mget Nodus3.2.2Q.sit.hqx? "n" {or "y", your choice}... ftp> "quit" 221 Goodbye.
VIII-138 Index VIII. INDEX 1 time axis range 71 3-dim coordinates 39, 59, 84, 87, 89, 90 50% Reduction 61, 63 About Nodus 55 accuracy 89-91 activation activation factor 20, 21, 26, 37, 44-46, 99-101 all comps 50-51 Allow editing 70 alpha function 23-24, 97 amplitude 75 Apple Menu 55 Automatic compartment names 59 Automatic loading/saving 29, 63 Automatic Saving 29, 62, 63 backup 35 background 7, 63 base amount 22, 94 Beep when finished 63 Before closing definition file 64 Before closing simulation file 64 binary branch 17 block 79 Block Ionic Currents 48, 79 branch connection 16-19, 39, 64, 89-91, 106, 111 branch connection icon 87 cable parameters 11-14, 36, 83, 84, 91 camera lucida 13 channel 20-21, 99 characteristic potential 22 check mark 28 Clear 62 Clear all synaptic events 46, 57 Clear experiments & text output 44, 57 Close 42, 60, 116 Close All 60, 116 Close All Graphs 60 Close simulation 42, 65, 116 CM 11 Comment 37, 40 compartment 10-15, 19, 33, 37-39, 49, 82-88, 91 compartment definition window 33, 84-88, 106 compartment labels 38 compartment name 50, 59, 64, 84 compartment number 37, 49, 86, 84-89 compartment preselection popup menu 49, 50-51, 64, 77 compartment selection popup menu 49, 50, 64, 75, 77, 82, 85 compartment shape 85 compartment size 84, 89-91 compartmental model 11, 12 compilation 44 Compile from 43 computation speed 14, 25, 46-47, 65, 90-91, 94 concentration 26, 48 conductance 20, 23, 79, 87, 97, 101 conductance blocking factor 49 conductance definition file 29, 32, 35-37, 58-59, 92, 97, 115 conductance definition window 38, 99-100, 107 conductance equation 99-100 Conductance menu 99-102 Conductance Plot window 60, 62, 99-101, 107 conductance popup menu 97 Configure Plots 41, 48, 64, 66, 70-71, 104 connection 16-17, 24, 86-87 constant transmitter release 24, 95 coordinate 84, 87 Copy 62 Copyright 2 cross connection 17 crosshair cursor 65-66 CS A Current 110, 113 CS Delayed Rectifier 113 CS Fast Na Current 113 Current Clamp 48, 74-76, 105 current display 75 current injection 26, 75 Current plots 63 Cut 62 cyclical 75 cylinder 11-13, 85 cytoplasmic resistance 11, 15 database compilation 43 Default to resting potential 63 Default to set potential 63 Default to values in memory 63 definition window 61, 64 delay time 24, 82 Delete all currents 74 Delete all firing times 78 Delete voltage clamps 76 dendrite 37 diameter 11, 18-19, 89-90 differential equation 12
Index VIII-139 double precision 24, 36 Double size simulation plots 61, 63 Draw axes 71 dual exponential function 23-24, 97 Duplicate 35, 93, 96, 98 dynamic equilibrium 44, 46 E 11 Edit axis 70 Edit Menu 62-64 electronic circuit 11, 24 electrotonic length 13-14, 47, 86, 90-91 empty selection 49 equation 36, 38 equilibrium 44 equivalent circuit 12, 15, 21, 24 equivalent cylinder 15 error message 51-53 estimated error 66 Euler method 24-25, 47 excitable compartment 33 excitable membrane 9, 20, 33 experiment 50 experiment commands 41, 48, 74-79 experimental data 36-37 experiments 47, 57, 74-79 exponential peeling method 15 factor 22 Fehlberg method 25, 47, 66, 69 File format 59, 61 file hierarchy 30, 37 file ID-number 30 file links 30 File menu 28, 55-61 file name 30, 33, 38 file structure 28, 30, 37 Finder 7, 30 firing 24 fixed time step 69 font 63 foreground 46 fuse 89, 91 Fuse Compartments 37, 40, 88-89, 91 gate 20 H 20, 99 Genesis 61, 118 Gmax 21, 23, 64, 93 Go to Compartment 88, 116 graphics window 61-63, 100 grow box 65 hardware 7, 46 heap memory 7, 55 HH Delayed Rectifier 112 HH Fast Na Current 112 Hide 60 Hide All 60 High multifinder priority 46-47, 63, 105 Hodgkin Huxley equation 20, -2322 hybrid method 25, 47, 66, 69 ID-number 30, 61 import format 37, 118-119 Import Neuron 37, 38, 59, 61, 85, 118-119 Import Simulation 59 in ms/cm2 64 inactivation 20-21, 37, 101 information 2 inheritance 44 initial values 44, 57, 63 injected current 49, 74-76 injection list 75 input resistance 13, 14 integration commands 42 integration error 26 integration method 24, 25, 68-70 Integration Settings 42, 66, 68-70, 104 integration speed 14, 25, 46-47, 65, 90-91, 94 invertebrate 17, 106, 120 ion concentration 10 ionic currents 10, 21, 33-38, 64, 87, 90, 92-94, 106 Ionic Currents 92-94 italicized 28, 42, 67 Keep 3-dimensional coordinates 59 Kill 60 λ 12 Ladder 61, 118 Leak 93 leak conductance 21, 93 length 11, 19, 89-90 linear cable theory 11 link 30, 32, 81, 86 local name 49, 81 lump 15 M 20, 99 Macintosh 5, 7, 46 major time step 25-26, 47, 69, 73, 75, 79 Make legend 73 mathematics 9 maximum 115 Maximum [ ] 69 maximum conductance 21, 23, 93, 97 maximum error 25 Maximum V 66, 69 Measure Window 28, 65, 66, 79, 80 measuring 65, 66, 80 membrane capacitance 11, 14, 86 membrane potential 26 membrane resistance 11, 14, 86 membrane surface 13-14, 19 membrane voltage 94
VIII-140 Index memory 7, 30, 43, 48, 55, 115 menu bar 27 minor time step 25-26, 68, 69 model 36, 41, 43, 51 model parameters 9, 14 morphology 10, 13, 14, 37, 40 MultiFinder 7, 46 Multiple use of subdefinitions 35, 64, 93, 95, 98 network 24, 36 network definition file 29, 32, 40, 58, 80, 115 network definition window 40, 80-81 Network menu 80-82 Neuron 36 neuron definition file 29, 32-35, 39, 40, 58-59, 83, 106, 108, 115 neuron definition window 83-84, 87, 88 Neuron Diagram 60, 62 Neuron menu 28, 83-98, 116 neuron model 34, 37, 91 neuron selection popup menu 49, 77, 81, 82 New 56-57 New Conductance 37, 57 New Network 40, 57 New Neuron 38, 57 New Simulation 41, 43-46, 51, 53, 56-57, 63, 116 Next Compartment 39, 88, 106 Next value type 73 NINCDS 118 node connection 16-19, 87-89, 91, 106, 111 Nodus 2 5, 7, 35 Nodus Preferences 7, 29, 63 Nodus Resume file 29, 62 not used 49 ns 64 Number of compartments 39, 47, 83 Only node connections 64 Open 32, 35, 57-59, 116 Open all linked files 33, 58 Open Conductance 58 Open Network 58, 109 Open Neuron 58, 106, 108 Open Simulation 41, 42, 44, 45, 46, 57 Optimize Model 37, 39, 40, 87, 90-91 option key 28, 116 original model 40 output 29, 48, 49 Output at major time step 73 Output maxima/minima 73 Oxford 119 Page Setup 61 paper 2, 120 parameter selection 51-53, 70-71, 74, 77-79 parameter selection popup menu 41, 48-50 parameter selection popup row 48-50 parameters 37, 88 parent branch 18, 19 parent compartment 17, 86-87, 106 passive compartment 11, 33 passive membrane 9, 13-14, 87 passive membrane model 14 Paste 62 Pause 42, 68 period 76 Plot (In)Activation 38, 101 plot buffer 72 Plot Conductance 100 Plot Rate Factors 101 Plot Scales 100-101 Plot Time Constants 101 Plot Window 28, 57, 60, 62, 64-65, 72, 79, 80 popup menus 28, 49 popup row 48-50, 70-71, 74, 77-79 postsynaptic compartment 33 postsynaptic conductance 23-24 postsynaptic current 23-24 postsynaptic neuron 82 postsynaptic site 23-24, 34, 36, 87 potential threshold 94 Preferences 35, 41, 44, 46, 47, 50, 60, 61, 62-64 preferences file 7, 29 presynaptic 22, 24 presynaptic compartment 33 presynaptic neuron 82 presynaptic potential 22, 24 presynaptic site 24, 35-36, 87 Previous Compartment 88 Previous value type 73 Print 61, 63 Print font 61, 63 process 22-23, 24 Quit 62 Quit when finished 63 ramp current 75 range preselection popup menu 51 rate factor 20, 99, 101 raw parameter selection numbers 52 reduced model 14, 15-19 Reference 2, 120 relative error 26, 66, 70 Repetitive sweeps 71, 76 Reset simulation numbers 63 resting membrane potential 12 Resting potential 46 RI 11 RM 11
Index VIII-141 RN 13 Run 41, 42, 64, 68, 104 run cursor 66, 68 Runge-Kutta method 24, 25 Save 30, 35, 61 Save As 30, 32-33, 35, 61 Scale Sizes 13, 87, 91-92 scaling factor 13, 84 scroll bar 65 Select by structure type 50-51, 64 Select only named comparts 59 selection of simulation parameters 38 selection popup row 48-50, 52, 64 Separated by TABs 73 Set Connections 40 SF 13 shift key 28, 88, 116 Show all compartments 75 Show only named comparts 50, 64, 85 shrinkage 13, 14, 91 shut off time 47, 69 side-branch 17-19, 111 simulation data file 29, 32-33, 36, 44-45, 57, 103, 115 simulation database 29, 36, 41-43, 48, 51, 53, 56, 62, 64, 70, 103, 116 simulation database windows 28 simulation length 69, 78 Simulation menu 28, 41, 42, 66-79, 104 simulation number 41, 63, 103 simulation parameters 44, 47-49, 56, 70, 81, 85 simulation plot window 64, 103 simulation run file 36 simulation speed 14, 25, 46-47, 65, 90-91, 94 simulation time 68 single precision 36 singular point 20 sinusoidal current 75 software 5, 122 soma 37 source 43 source model definition file 32, 41, 43, 48, 52-53, 58 space constant 12 sphere 11, 13, 85 split 90, 91 Split Compartment 37, 40, 87, 89-90, 91 spreadsheet 29, 37 Squid Giant Axon 112 Squid Giant Axon Demo 112 state 99 Status Window 28, 65, 80 steady current 75 steady state conductance 100 step 76 stiffness 25-26, 46-47 Structure type 38, 39, 50, 85 structure type preselection popup menu 50, 64 subdefinition 34, 35, 38-39, 64, 84, 90, 92-98, 106 subdefinition dialog window 34 subdefinition popup menu 49, 87 subdefinition selection 33 subdefinition selection popup menu 34, 92-98 subdefinitions 33, 87, 89 submenu 27, 117 synapse 22, 47 synaptic conductance 26 synaptic current 34, 82, 90, 98 Synaptic Currents 33, 34, 38, 64, 87, 96-98, 108 synaptic event 46 synaptic firing time 24, 79 Synaptic Firing Times 48, 50, 78-79 synaptic shut off time 47, 69 synaptic switch off time 24 System 7, 46 table 25, 47, 69 template 43 Test 7 Network 109 Test 7 Network Demo 108 test model 38 Test-cell 1 110 Test-cell 1 Clamp 110 Test-cell 2 106 Test-cell 2 Demo 103 Test-cell 3 110 Test-cell 4 111 Test-cell 5 111 Test-cell 6 108 Test-cell 6 Demo 1 107 Test-cell 7 109 text attributes 5 text file 29, 59, 61, 72-73 Text Output 41, 48, 66, 69, 72-74 text output file 29, 72 Threshold 73 threshold potential 22, 94 tie 89, 106 Tile windows 63 time constant 12, 14, 86, 101 time sharing 46 time to peak 97 Time Window 28, 65, 80, 103 τ m 12 transmitter amount 79, 94
VIII-142 Index transmitter release 22, 24, 33, 35, 87, 94-96 Transmitter Release 38, 94-96 transmitter release site 82 tree format 17, 39, 85 tree model option 85, 87 triangular current, 75 Trigger all output 73 tuning 39, 40 type 1 synapse 23-24, 78, 94, 95 type 2 synapse 23-24, 95, 97 Undo 62 unit popup menu 71, 74 Update all links 30, 61 Update all links in memory 61 Update Plots 71 Value 70 value type selection popup menu 48, 67 Values in memory 45 variable membrane resistance 15 variable time step 24, 69 variable transmitter release 24, 26, 95 vertebrate 17, 121 View/Edit Parameters 46, 48, 66, 70 virtual memory 46 voltage clamp 22, 26, 38, 48, 49, 63, 71, 76-78 voltage clamp cycle 76 voltage clamp length 76 Voltage Clamp 76-78 voltage dependent 20 voltage range 47 Warnings 60, 62, 64 warranty 2 weight factor 16-17, 19, 39, 64, 87, 106 white noise current 75 window commands 42 V/ t 70