ABI CellML Meeting Minutes, 28th July 2010

Present: Dougal Cowan, David "Andre" Nickerson, Randall Britten, Justin Marsh, Andrew Miller, Peter Hunter, Tommy Yu

Last week's action items:

1) Tommy - edit the authorship information in the application note using MS Word.

  • Done.

This week's agenda:

1) Peter Hunter: support for running models through CUDA (tracker item 2644)

Discussion:

  • Peter: Marco Viceconti wants to run CellML models on CUDA, using many CellML integrators in parallel to perform sensitivity analysis. Andrew McCulloch's group is also interested in CellML on CUDA.
  • Peter: Chris has a Masters student working on CUDA. It is probably best to let Chris and the other groups collaborate.
  • Randall: Is there concern about publications, or keeping the work under wraps?
  • Andrew: There are a number of different things CUDA could be used for, and it is likely each will need to be implemented separately. Examples include sensitivity analyses, spatial / tissue models, and speeding up individual simulations using parallel vector computations.

2) CellML core specs - update

Progress:

  • Randall: Mark Hanna is doing a 4th year project on delayed ODEs. He is looking at methods for solving delayed ODEs and ways of representing them.

3) CellML API - update

Progress:

  • Andrew: Has changed the units validation to be stricter.
  • Andrew: The CellML2C test program now allows testing with named variables in IDA mode.
  • Andrew: The developer documentation associated with the API has been improved, and now also provides basic instructions on how to use the test programs.
  • Andrew: Has added support for cloning a model and all imports that have been loaded; this is functionality OpenCell had, which is now available through the API.
  • Andrew: Added support for working with extension attributes from the CellML API. Is still working on unit tests.

Planning:

  • It was decided that it is acceptable practice to add new functionality to the CellML API without community consultation provided existing methods are not impacted.

4) OpenCell development

Progress:

  • Peter: After discussions with Randall, Alan, Justin and Andrew about how OpenCell will be developed, a decision has been made to move the leadership of the OpenCell development to Oxford, with Alan Garny heading the development. Making sure that the development of OpenCell is part of an international open-source project is a major focus that we wish to encourage.
  • Peter: The Auckland group will focus on the APIs (CellML and FieldML) and web services (eg. PMR2).
  • Peter: There will be increasing pressure to provide good APIs and web services, for example for the Nature Physiome Gateway, which could provide a way to cite model building and curation.
  • Justin: Has been trying to find the right working version of the Saucerman model on Geoff Nunns' VM. Two potential working models require testing.

Discussion:

  • There was a discussion about citations and credit for work on building and curating CellML models.

Planning:

  • Andrew: Due to the plans to rewrite OpenCell, there needs to be good support for whatever language Alan decides to write the release in.
  • Andrew: It also makes sense to move features from OpenCell to the CellML API sooner rather than later, so the current code in OpenCell doesn't need to be rewritten for the new platform.

Planning:

  • Randall: There will be a final release of OpenCell using the XULrunner codebase.

5) Metadata specifications (general, citations, graphing, simulation, other) - update

  • No update.

6) Repository contributions - update

Progress:

  • Hanne: Finished creating sessions for the metabolism section and is now working on the signal transduction section
  • Catherine: Continuing to curate models in the metabolism section and has been able to get the Hynne et al. 2001 model of glycolysis working by comparing it with the curated SBML model in the BioModels database.
  • Peter: Catherine is working on Liver models for the German virtual liver project. This is a major 5-year project that aims to produce a complete multi-level model of the liver.

7) PMR2 development - update

Progress:

  • Tommy: Finishing up the application note.

Discussion:

  • There was a discussion about short and custom unique URLs to exposures or workspaces.
  • Randall: we need a disambiguation page to handle collisions of shortened URLs.

Planning:

  • It was agreed that we need to establish consensus on the best way to handle names for exposures.

8) Time permitting: Updates or additional items from volunteering attendees

Discussion:

  • Andre: Catherine is very keen on a bulk fix for the e-notation and xml:base problems in the repository. We should at the very least run a script which finds all the problems and generates a list Catherine can use to fix the models. Once this is done we can assess the need for a script that fixes the problems.

Action item 1: Justin - Look into a script to find models that require fixing.