How to create and interpret the predictive analysis of a compound
Platform with suite of tools Predict & understand biological effects of small molecules & compounds Predict targets and metabolites, potential disease and toxic indications and evaluate potential mechanism of action of your drug Structural Analysis Tools Functional Analysis Tools Identify biological processes, toxicities, diseases etc statistically associated with list of compounds, metabolites, targets Optional: Supplement with toxicogenomics data.
Several options to start workflow 2. Select Predict Compound Activity (Metadrug) from search results OR Compound information page OR Toxic agent information page 1. Search for compound
Can upload multiple compounds in one file (note limitations and requirements apply). To include all compounds / metabolites in one report select include all compounds in SDF file into one report. 2. From start page start a Compound Activity workflow and Upload compound from CDX, SDF or Mol file type or draw structure
Step 1of 2 Name the file to be saved in your data manager and select to turn metabolite prediction on or off with Prioritization (sorts metabolites into major, conjugated and minor) and if to include second pass metabolites Browse and select reaction types to include. Uses > 150 metabolic rules describing common metabolic reactions, categorized according to the particular type of chemical transformation (e.g., aromatic hydroxylation or ester hydrolysis).
Step 2 of 2 Set the Tanimoto similarity threshold in the Threshold box (from 0.4 to 1, higher value corresponds to more exact matching). The similarity to the input structure is inspected for each chemical compound in MetaCore database (>677,000). Select QSAR models to calculate ; click down arrow to view / change default threshold settings. Option to incorporate your own QSAR model Select Additional options if required
Custom QSAR Models QSAR models can be extended with user-built Custom QSAR models. To build your own QSAR models with proprietary data a use wizard from start page Every QSAR model (custom or built-in) can be linked to biology through canonical pathways and networks
From Start page remember to REFRESH browser and your report is now under Structures in My data File From this report - work your way through each section. Option to change the Similarity search and recalculate and export the full report to excel. This maybe useful to increase number of possible targets ie: 0.5. Static part of report. Options to change similar compounds and possible targets that are used in enrichment analysis Dynamic part of report.
Explore the predicted metabolites of your compound -Export properties as sdf and then predict activity of these metabolites by building another Metadrug report -Export properties (and QSAR models if selected) to excel and filter -Include all metabolites in one report so only one file is produced The prioritization of metabolites is based on a score representing the Occurrence rate (OC). Ratio of the occurrence of a particular metabolic reaction to the total number of metabolic reactions in the database. Assigned to predicted metabolites as a negative log value (logoc).
Predicted metabolites Over 80 rules to predict likely reactive metabolites such as quinones, aromatic and hydroxyl amines, acyl glucuronides, acyl halides, epoxides, thiophenes, furans, phenoxyl radicals, phenols, and aniline radicals. Molecules with reactive groups are marked and highlighted. The prioritization of metabolites is based on a score representing the Occurrence rate (OC). Ratio of the occurrence of a particular metabolic reaction to the total number of metabolic reactions in the database. Assigned to predicted metabolites as a negative log value (logoc). The larger the score, the higher is the frequency of the metabolic reaction in our database.
Expand each section to show predicted values for QSAR models. The details of each model are provided along with calculated values. The top value is the QSAR value while the lower number in brackets is the Tanimoto prioritisation (TP) value (how similar the input compound is compared to the training set used for that QSAR model) Green Values mean the compound is within the threshold values of the QSAR model or for the TP value that the compound is >50% similar to the training set used for the specific QSAR model Red values indicate the compound is NOT within the threshold values of the QSAR model or for the TP value that the compound is <50% similar to the training set used for the specific QSAR model The Green M indicates that this QSAR model can be calculated for the metabolites as well as input compound if selected in the additional options section. The thresholds for each QSAR model are set at default values but can be changed during STEP 2 of the Data Analysis Wizard.
QSAR models for Therapeutic Activity The predicted QSAR values greater than threshold indicate the potential therapeutic activity and are colored green. The quality of the model is evaluated in terms of specificity, sensitivity, accuracy and MCC. For the Thomson Reuters QSAR models the training sets can also be exported and viewed. Colours are a guide only. All values and thresholds should be considered by the user and their level of restriction. ie: if have complex / unique compound may want to relax the TP value where values <50 maybe considered
QSAR models for Toxic End points The predicted QSAR values NOT within the thresholds indicate the potential toxicity activity and are coloured red (think red is BAD Green is Good). The exception to the rule is the skin sens, EC3 QSAR model. Where values >10 indicate a potential sensitizer and are shown in RED. More specific details of the QSAR models can be found in the reference
Select similar compounds to use in target prediction Identify similar compounds target prediction (Guilt by association) - similar pharmacological or toxicological properties - Grouping or read-across - Infer activities of uncharacterised compounds of similar structure Hover over compound name to view structure and select those to use to predict potential targets for input compound
Possible targets Export list to excel or as a new experiments Group targets by Protein class Select which targets to be used in dynamic enrichment analysis
Dynamic part of report: Based on all previous selections perform enrichment analysis in 9 different ontologies INTERACTIVE: Click on map / network to visualise possible targets, potential compound affected processes and how your compound is working Green number number of network objects from possible target list. Red number number of network objects in map / network P-value probability of a random intersection between set of target Ids and those in map / network / ontology Differences observed from enrichment with genes / data 1) No of targets typically similar than in gene list. This limits applicability of distribution of p values 2) Drug targets usually represent highly connected protein (hubs) involved in many pathways and processes. Limits specificity of distribution
Review Pathway Maps Formulate a hypothesis of how your compound works. Identify biological processes, toxicities, diseases etc statistically associated with list of targets predicted by compound activity report. Overlay more visualisation options to display information on Network objects are associated with Toxic Pathologies or Diseases, are drug targets or even overlay your own experimental list / data.
OPTIONAL: Use targets of input compound to perform enrichment analysis of further ontologies Disease (by biomarkers) ontology is not covered in report but enrichment analysis for this ontology is easy by exporting the targets and performing a one-click analysis Export targets from Metadrug report of input molecule.
OPTIONAL: Use targets of (1) input compound and (2) predicted metabolites for further enrichment analysis Export targets from Metadrug reports of input molecule and combined metabolites as experiments. Activate and use one-click analysis to explore enrichment analysis for single ontologies or use the workflow and reports to analyse single experiment