ABI CellML Meeting Minutes, 28th April 2010

Present: Andrew Miller, Dougal Cowan, Tommy Yu, Catherine Lloyd, Poul Nielsen, David "Andre" Nickerson,  Geoff Nunns, Randall Britten, Justin Marsh

Apologies: Peter Hunter

This week's agenda:

1) (Randall) Update on the PMR2 papers.

  • Randall commented that we want to ensure that the relevant people are included as authors.
  • Poul commented that this should be as inclusive as possible.
  • There was a discussion about the content of the ICSB abstract.
  • Andre said that he may write a PMR2 abstract for ICSB, and also one about CellML 1.1 best practice.
  • There was a discussion about Andrew's abstract on PMR2.

2) (Andre) is this something we should participate in? http://bioinformatics.oxfordjournals.org/cgi/content/short/26/8/991

  • Poul said that it would be sensible to have a presence in this group (www.biocuration.org).
  • Andre said that the main thing was to be a part of the community.

Action item 1) Catherine to follow up on membership of biocuration.org.

3) PMR2 development - update

  • Tommy has staged the alpha on Amazon, including recent bug fixes and updates. He needs to re-create the FieldML demo exposure.
  • There was a discussion about testing the PMR2 alpha.
  • Tommy did some work on re-naming variants (legacy PMR1 content), as this was using the wrong field.

4) Repository contributions - update

  • Geoff reported that he has been working on a (?) pentose phosphate model, and has been working generally on metabolism models, converting many from SBML.
  • Catherine reported that she and Geoff are going through the repository section by section and are currently working on the metabolism section.
  • Catherine has been fixing the names of models to help with distinguishing them from one another.
  • Andrew asked how well his SBML converter is working, and Geoff said that a few things require manual fixing. There was a discussion about the value of further work on this converter to get it to produce better output.
  • There was a discussion about the interoperability grant, and working on interoperability issues.
  • Andrew said that a great deal of benefit could be derived from improving converters, and Poul agreed.
  • Poul noted that there are other potential sources of funding for work on interoperability.

5) CellML core specs - update

  • No update.

6) OpenCell development - Qt/C++ verses JVM/Scala, and what to do next?

  • Justin has now included a graphing demo in his prototype. He can now build, but not package or run the project without Eclipse.
  • Randall suggested that Justin give a demo of the prototype after the meeting. Alan will need to see a demo as well, possibly via EVO video conferencing.
  • Andre said that another possibility (other than Qt/C++, JVM/Scala, or XULRunner) is to abandon OpenCell altogether, and work on new tools.
  • There was a discussion about tools and alternative options to OpenCell.
  • There was a discussion about what tools, features and so on are desired or required for working with CellML.

  • Randall suggested the first project to work on under the new platform could be a simple stand-alone metadata editor.

  • Randall asked about SVG support in JVM/Scala. Justin said there is SVG support, but integrating JavaScript for sessions is not possible. However, the way sessions are done needs a redesign anyway.

  • Poul asked what the criteria were for the selection.  Andrew and Justin outlined the main factors: usability, productivity, maintainability. Some factors influencing these: availability of libraries and components, ease of integration, ease of multi-platform support, code readability. Andrew feels that Qt/C++ is slightly less preferable to status quo, and his preliminary view is that JVM/Scala/Eclipse is preferable to status quo. Randall feels strongly that we should avoid C++ for this type of application, and favours JVM/Scala/Eclipse.

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

  • Catherine emphasised the need for a spec, and a metadata editor. She mentioned that SEDML is going to be a big part of the SBML-Biomodels Hackathon in Seattle.

8) CellML API - update

  • Andrew has been looking at Python bindings, which don't seem too difficult.
  • Randall has helped Geoff to try IDA for the Saucerman model, but this doesn't work yet. He mentioned that it could be useful to generate the residual method in MATLAB from the CellML, to see if it is a problem with the CellML code.
  • Randall said it would be good to get the Saucerman model and Jonna's Guyton model going under IDA.

  • Andrew said that initial condition hints specified in metadata are not being used yet for the IDA code generation. Randall said that implementing this is very likely to address the current problems with using IDA for the Saucerman model. Randall said that about a year ago he adapted the original Saucerman MATLAB code to work with IDA in MATLAB, and the results matched the original results virtually perfectly, which indicates that IDA would handle the Saucerman model. However, this did show that the initial conditions were an important factor. Randall also did a small amount of testing of IDA on a test DAE that he created that has two alternate trajectories, and IDA appeared to handle this better than any of the other solvers he has tried (MATLAB; decoupled approach in CellML-API CVODE/levmar; Mathematica).

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

  • There was a discussion about Mark Hanna's project on delayed differential equations in CellML.
  • Catherine's remote attendance and involvement in meetings while she is away was discussed.
  • A decision was made to add a DRAFT label to the meeting minutes before they are published.