The UK INCF Node and the CARMEN project Presenter: Leslie Smith l.s.smith@cs.stir.ac.uk
Content Events and activities at UK INCF Node Recent Forthcoming The CARMEN project History, current status, ways forward. INCF Nodes meeting Warsaw April 2015 2
Recent events Integrated Systems Neuroscience Workshop 2015 23-24 March 2015, Manchester University A workshop to demonstrate the power and potential in state-of-the-art integrated systems and computational approaches to key neural circuits. Themes Cortical circuits Basal ganglia Attention Large-scale recording of circuits http://www.isn2015.ls.manchester.ac.uk INCF Nodes meeting Warsaw April 2015 3
Recent Activities Stephen Eglen took part in the Academy of Medical Sciences Symposium 1-2 April, London Reproducibility and reliability of biomedical research JB Poline presented A report will be published There are ongoing discussions with Wellcome Trust, MRC on enabling re-use of both data and services/workflows, relevant to bioscience datasets. HBP-related Neuroinformatics: Bruce Graham took part in a meeting on open-source community-based models of hippocampus CA1 31 March -1 April Leslie Smith took part in Open Data e-infrastructures for neuroscience 5 March Thomas Novotny and Marc de Kamps presented at HBP Theory and Neuromorphic Hardware workshop 2-3 April INCF Nodes meeting Warsaw April 2015 4
UK Node Website The website is now hosted at Stirling University Same URL: http://www.neuroinformatics.org.uk Stephen Eglen has organised a list of UK Node member publications http://tinyurl.com/uk-incf-papers INCF Nodes meeting Warsaw April 2015 5
Looking Forward UK will host INCF 2016 annual meeting In Reading, near London. Ingo Bojak is local organiser A proposal for a Neuroinformatics Symposium has been put forward to FENS 2016. 6 INCF Nodes meeting Warsaw April 2015 6
CARMEN Code analysis, repository and modelling for e-neuroscience) is a project set up to provide a portal-based system allowing Storage of time-series data (electrophysiological datasets) Analysis of these datasets by services and workflows at the portal Visualisation of datasets The concept was that researchers would use the portal to share data and analysis techniques With collaborators (a colaboratory) With the research community in general (making them public) Funding for the project started in October 2006. INCF Nodes meeting Warsaw April 2015 7
CARMEN architecture INCF Nodes meeting Warsaw April 2015 8
CARMEN history UK EPSRC funding 2006-2010 UK BBSRC follow-on (tools and techniques) funding 2010-2015 but it s run out of time and money now. The portal is there http://portal.carmen.org.uk INCF Nodes meeting Warsaw April 2015 9
INCF Nodes meeting Warsaw April 2015 10
CARMEN: very much in use INCF Nodes meeting Warsaw April 2015 11
Where now for CARMEN? It s up and running, and being used: it s almost full But more disk space is being found. But: to expand it, or to continue it properly needs another major grant The equipment running it is getting old Some of the design is perhaps out of date Internal parallellism is missing Funding has run out INCF Nodes meeting Warsaw April 2015 12
Ways forward 1. Give up on it It was a good idea at the time, but actually It s too hard Neuroscientists don t really want this sort of thing sufficiently Infrastructure is too hard to keep running 2. Mothball it Keep it running (get some more disk space first!), and hope that the University where it is sited will let it be looked after along with other servers 3. Look to expand it in an appropriate fashion Get more funding, but not just to continue it as it is INCF Nodes meeting Warsaw April 2015 13
Ways forward 1. Give up on it It was a good idea at the time, but actually It s too hard Neuroscientists don t really want this sort of thing sufficiently Infrastructure is too hard to keep running 2. Mothball it Keep it running (get some more disk space first!), and hope that the University where it is sited will let it be looked after along with other servers 3. Look to expand it in an appropriate fashion Get more funding, but not just to continue it as it is INCF Nodes meeting Warsaw April 2015 14
Client pull or technology push? INCF Nodes meeting Warsaw April 2015 15
Client pull: the users Who are the target users? Clinical neurologists and neuroscientists Epilepsy, traumatic injury, Parkinsonism, Neuropharmacologists Assessing effectiveness of neuroactive pharmaceuticals Research neuroscientists In Universities and hospitals etc Neuromodellers Data to constrain and test models Educators Training the next generation of neuroscientists INCF Nodes meeting Warsaw April 2015 16
Making the system user-focussed What do prospective users want? What do they need? What is the problem the system is trying to solve? What will they actually use As opposed to what they say they might use? How can the system be made attractive and straightforward enough for neuroscientists to use? What are the issues that discourage users? INCF Nodes meeting Warsaw April 2015 17
Technology push What technologies might be helpful? Note that neuroscientists don t want bleeding edge technology in their support systems As opposed to their scientific systems! Handling large datasets Remotely visualising large datasets Parallelism At the user level (multiple simultaneous users) At the processing level (e.g. multiple datasets, or parameter searching) Search technologies Searching metadata, services, workflows. INCF Nodes meeting Warsaw April 2015 18
What do the users want: 1 Ease of use Not simply a question of good UI design! Data upload requirements: Single data entry Preferably automated, from existing equipment Single metadata entry Preferably from same equipment and at same time as data E-Lab book Or at worst, entered into a single system once INCF Nodes meeting Warsaw April 2015 19
What do the users want: 2 Data Discovery Data must be easy to find Implies accurate metadata Searchable metadata: based on a shared ontology Data analysis issues Ability to set up analysis workflows And use them on multiple datasets fropm many labs Implies data format compatibility Data visualisation issues Straightforward remote visualisation INCF Nodes meeting Warsaw April 2015 20
What organisations are relevant? INCF? Strategic Objectives 1. Partner with international stakeholders to promote and prioritise neuroinformatics at global, national and local levels Human Brain Project?. What is their relationship to the EU Horizon 2020 infrastructure project? Research Data Alliance?. https://rd-alliance.org GÉANT. http://www.geant.net/pages/default.aspx EUDAT INCF Nodes meeting Warsaw April 2015 21
And which funding agencies? European Research Infrastructures, including e- Infrastructures (Horizon 2020)? http://ec.europa.eu/programmes/horizon2020/en/h202 0-section/european-research-infrastructuresincluding-e-infrastructures Are there others? COST? Can we link out to funders outside Europe? Plus national research councils, perhaps charities as well. INCF Nodes meeting Warsaw April 2015 22
Which agencies to target? Neuroscience infrastructure Enabling data and analysis technique sharing (I don t need to explain why that is important to this audience, I hope) EU in Europe, NIH, Kavli in US, elsewhere? INCF Nodes meeting Warsaw April 2015 23
What shape might a project have? Infrastructure Area? Scientifically speaking (But also geographically!) Aims? Leadership? Relation to other stakeholders? Intended impact? INCF Nodes meeting Warsaw April 2015 24
Scientific Area? Which elements of neuroscience? CARMEN time-series data in Neuroscience? EEG, ECoG, Electrophysiology, time-series imaging, A large and growing area Large volumes of data Multiple application areas Clinical, neuro-pharmacological, behavioural, purely scientific, Lots of interesting problems and applications Data formats, spiking and LFPs, BCI, Eplilepsy, Parkinsonism, Is this too narrow? Not a large enough community? Is this too broad? Making it difficult to provide specialist tools and appropriate infrastructure? INCF Nodes meeting Warsaw April 2015 25
Geographical Area? Europe? The US? The World? Really a question of who will fund such an infrastructure project. Current likely candidate: EU Horizon 2020 Infrastructure funds, NIH, others? Can INCF help to locate appropriate funding streams? INCF Nodes meeting Warsaw April 2015 26
Leadership Should there be a lead organisation INCF? In Europe: HBP? In US: BRAIN Initiative? or the UK Node? Still needs a group of people to organise it And probably organise a large consortium Other infrastructure projects tend to be large Many collaborating groups INCF Nodes meeting Warsaw April 2015 27
Relation to other stakeholders Specifically: HBP, ELIXIR, EU Infrastructure INCF RDA GÉANT Can we get them all on board? Including funding agencies H2020, NSF, NIH, plus all the national ones Who might support this? INCF Nodes meeting Warsaw April 2015 28
Intended impact Enabling effective searching and sharing Of data Of metadata Of processing tools/services/workflows Cross-analysis and re-analysis of datasets Re-using expensive datasets with different tools Comparing results from many experiments in different labs Comparison and re-evaluation of tools. Analysing dataset with new and different tools INCF Nodes meeting Warsaw April 2015 29
Making it work Bringing the tools to the data Because the data is too large to move easily But making this effective and easy to use is difficult Needs effective portal techniques Designed for the users, quality of user experience High bandwidth links for visualisation Linked to other resources (e.g. NIF, ontologies, ) Preferably free at the point of usage But sustainability requires that someone is paying INCF Nodes meeting Warsaw April 2015 30
Concluding remarks Everyone wants infrastructure But they disagree about how to organise it And who should pay for it Large data stores, parallel computers, fast links make e-infrastructure possible But they only solve some of the issues Not ease-of use, nor conservative attitudes One way forward is educating the next generation in Neuroinformatics Though this is rather a slow solution INCF Nodes meeting Warsaw April 2015 31