ABI CellML Meeting Minutes, 18th of May, 2011

Present: Dougal Cowan, Andrew Miller, Tommy Yu, Randall Britten, David "Andre" Nickerson, Poul Nielsen

This week's agenda:

1) Representation of parameter uncertainty: SED-ML or CellML 1.2

Discussion:

  • Randall: the discussion of this at the CellML workshop was recorded.

Action item 1: Randall - get information about hosting of these recordings.

  • Randall: Peter wants to revise the decision at that was made at the CellML workshop, that the specification of parameter uncertainty should be deferred till after 1.2 (tracker item 2809). This is because we have funding (VPH Share) for the work, with the obligation that the CellML itself allows for the description of the probability distributions of uncertain parameters. Andrew will be working on this. This would mean including it in the list of blockers of tracker item 55 again. The rationale at the workshop was that handling stochastic variation has a lot of complexities, and that the VPH Share requirements are a special case that could be handled in the interim by extending and supporting SED-ML.
  • Andre: The list of tracker items listed under tracker item 55 represents community consensus on what is required for CellML 1.2. This was discussed, and it was concluded that we would interpret the inclusion under the CellML 1.2 tracker item more broadly, so that de facto contributions that will hopefully become accepted proposals are also listed. Andre: Stressed that it is important that we do not reinvent the wheel, and thus suggested that Andrew first consult the work that SBML have done on the SBML distributions package.
  • There was a discussion about whether these changes should become part of 1.2, a secondary spec, etc.
  • Andre: we need someone to become familiar with the SBML distribution package, and ideally use it or at least be compatible with it. 
  • There was a discussion about the simulation vs model issues to do with specifications.
  • There was a discussion about how the VPH Share grant affects these decisions.

2) CellML core specs - update

Discussion:

  • Poul: attended multicelluar modelling workshop. CellML or SBML are not currently able to represent these dynamic multicell models, although for some types of static cell number models CellML can be used.
  • Andrew: it is possible to encode these dynamic models with ModML, as long as there is a fixed upper limit on the number of cells.
  • There was a discussion about how to model these problems.
  • Andrew: ModML - the ModML seminar was well-attended.

3) CellML API - update

Progress:

  • Andrew: working on the build system to get builds that we can test. There is some documentation on the cellml.org site. Trying to get MinGW working with cross-compiling from Linux to Windows.

Planning:

  • Randall: Based on discussions with various people, this release will target the platforms that are already in the release candidates. After the release, Andrew will focus on the additional platforms, and make a release as soon as these are also successfully supported. The rationale is that delaying the release until the new platforms are supported would not result in their delivery any earlier anyway. Also, if bioeng46 can be upgraded to a 64bit OS, then we will upgrade it, and possibly cease support for 32bit MacOS binaries. If it cannot be upgraded we will consider getting a new Mac or using another Mac for 64bit builds.
  • There was a discussion about tracker item 2918, about whether these files should be included.
  • Andrew: the file could be included in this version with a note saying that it is non-standard.

4) PMR2 development - update

Progress:

  • Tommy: previous refactoring accidentally removed embedded workspace listings, these need to be added back.
  • Randall: David Bullivant is making progress on a Buildbot system.
  • Tommy: the OS upgrade has been delayed due to the RedHat license expiring. ITS are working on it.

5) Repository contributions - update

Progress:

  • Dougal: have been working with David Friel on his request that we fix the Friel 1995 model in the repository. David has fixed the CellML himself, based on his IgorPro implementation of the model.

Discussion:

  • Randall: should we be putting a CSV file of the results that provide a match to the figure in the paper in the workspace with the model as part of curation? This could be used later for numerical validation of the model.
  • There was an extensive discussion about this method of numerical curation, and other curation issues.
  • Randall: we need to be more careful about the wording of model status paragraphs, not to imply that there are problems with the original paper, since often the problems are just due to human error during the manual translation to CellML.

Planning:

  • There was an agreement that we need to move away from session files and towards SED-ML descriptions of simulation experiments, which should be separate from the models.
  • Randall: we can go ahead and start adding CSV files to workspaces now.