ABI CellML meeting minutes 2009-11-04

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

Apologies: Mike Cooling, Peter Hunter

1) (From two weeks ago) Dougal to write a guide on how to create session files with SVG diagrams.

Not done - James and Dougal to write some of this guide before James teaches Hanne.

1) James to add the offer of hosting to the site.

1) (Catherine) Do we want to apply to host a tutorial at ICSB 2010? The application deadline is Jan 15th

Randall said that we can apply to host a tutorial at this conference - if the application is successful, we can host for free.
Randall said that while funding needs to be organised, people can express their interest in attending.
The conference is in Edinburgh from the 10 to 15 October, 2010.
Andre mentioned that IEEE EMBS is worth looking at as well.
Poul said that Biobricks would be appropriate for ICSB.

2) CellML specs - update

3) Metadata specifications (general, citations, graphing, simulation, other)

James said that he would like to complete the draft so that Tommy can use it for the metadata editor in PMR2 0.3.
Randall mentioned that we need to prototype the use of MIASE in OpenCell - perhaps after the 0.7 release, we can write this into the 0.8 plan.

4) OpenCell development update

Action item 1) James to make a new tracker item about "broken sessions", unless existing TI is found.

Randall asked if there was an estimate of when OpenCell 0.7 can be released.
There was a discussion about problems (such as lack of a Mac version) that are holding back the 0.7 release.

Action item 2) Randall to investigate why there is no development Mac yet.

There was a discussion about bioeng46, and why it is still on MacOS X 10.4
Justin reported that there are still a number of bugs to fix, and he would like to migrate some major tracker items to the 0.8 release.
Randall said that we should not consider adding the TXT view until the OpenCell text editor is fixed.
Randall reported that he has made a minor fix to the MATLAB code generation, based on Andrew's suggestions; tracker item 2070.

5) CellML API - update

Andrew has fixed a reasonable number of bugs.
Andrew reported that support for definite integrals has been fixed.
Randall asked what would happen if non-C code generators had to generate code with these definite integrals.
Andrew said that whenever we change something like this, someone will have to update the code generators.
Randall said that we should add tracker items for fixing the code generators; create TIs that say something like "the following language generators don't support this feature".
Randall suggested that we could update the documentation to give details of features that don't work in some language exporters.
Andrew said that there is a lot of uncertainty around this; someone would need to test and fix each case.
There was a discussion about automated numerical testing; this could become part of the 0.8 plan.
Andrew is still working on fixing some other issues, such as memory leaks.

Action item 3) Andrew to create a tracker item about Valgrind testing.

Randall said that Hugh is working with IT to sort out a shared testing framework for a number of ABI software projects.

6) cellml.org update

James reported that the workshop 2010 pages and registration forms are now up.
A Plone form generator has been added to the site to make form creation much easier.
Catherine is working on the community section, creating case-study style content on groups using CellML.
We plan to create an assets/logos page.
We need a terms of use page for PMR2, particularly we need to give information on how to cite the diagrams. We will put up a citation template.

7) Repository contributions update

James contacted Nicolas Le Novère about MIRIAM; he said that the website is the same as the paper, with the addition of suggestions.
James has added the paper referenced in the API paper.
James has been showing Hanne how to do keywords, will get her creating sessions soon.
Poul expressed concern about our current ad-hoc keywords system; should be be using SBO terms or some other controlled vocabulary?
There was a discussion about the keywords, and alternative ontologies to use.
Poul said that we need to assign detailed terms to models so that these terms can be used in future to automatically categorise models into the keyword categories. It is easiest to annotate models against SBO as they are being created.
There was a discussion about the use of Saint, and whether to start annotating models in detail now.
Andrew said that it should be relatively easy to add SBO terms in RDF to annotate models.
Geoff has been working on the Saucerman model, and has the exported MATLAB code reproducing the graphs from the original MATLAB.
Randall said that this model is an example of where the OpenCell solver has numerical problems.
Randall said that he will try using different solvers in MATLAB to try and pin down the difference.
There was a discussion about solvers to try.
Andrew said that changing the solver is an easy fix if this turns out to be the difference.

8) PMR2 development - update

Tommy is finishing the foundation for the new exposure styles, and has started migrating old object types into the new structure.
Tommy has also been cleaning up unnecessary portlets.

Action item 4) Tommy to meet with Michael Backhaus and Roger Clayton to discuss design issues around exposures.