ABI Meeting Minutes 2008-04-16

Present: Tommy Yu, Catherine Lloyd, Andrew Miller, Justin Marsh, Poul Nielsen

Apologies: Randall Britten

Review of last week's action items:

  1. Catherine to send e-mail to arrange a meeting about the CellML website

Catherine has sent out an e-mail to a few people to organise a meeting to discuss the future of the CellML website. Poul indicated that he would also be interested in attending, so Catherine will forward this to him as well.

Agenda for this week:

Priorities for work on CellML tools

  • Andrew has suggested that we go over the priorities for the development of CellML tools.
  • Peter Hunter has expressed a desire to have solving of differential-algebraic models, as required for the Saucerman-McCulloch model, working as soon as possible.
  • The two approaches possible here are to design an algorithm to identify the systems which need to be solved (which doesn't appear to have an obvious efficient algorithm to solve exactly) or to support this by annotating systems of equations using metadata (which is conceptually unclean, and also has the problem of requiring RDF/XML processing at the API side, which would make an RDF/XML parser a prerequisite for this development).
  • Alan Garny has responded to the action item on the mailing list to note that the Kyoto group have a solution to this problem which doesn't require systems to be explicitly annotated, and they didn't like the idea of needing to annotate such systems.
  • Action Item 1: Andrew will contact Alan to get further information about this solution to see if it is applicable to us. He will also send out an e-mail to the CellML discussion mailing list to solicit opinions on this from the wider community.
  • Waiting for this feedback will likely take some time. Until then, Poul noted that working on the RDF/XML processor will be beneficial whichever approach we take. Justin suggested that Andrew work with him doing pair programming to implement this, so Justin is familiar with the solution we end up with. This will mean that Justin will be able to continue working on it if Andrew moves back on to implementing a non-metadata based solver later. There was agreement that this was a good way to move forward.
  • There was a brief discussion of whether Alan Garny might also be able to help out with some of these goals, and Justin explained his understanding of what Alan is currently doing with the YAPE prototype.

Assisting imports of modules using sub-modules (embedded workspaces)

  • Tommy has looked into the Mercurial submodule support, and explained that the extension which provides this doesn't work well with Zope. This is something he plans to eventually look into further.
  • Tommy's main concern now is the complexity of presenting the information in an intuitive way to the user, if users want to create forks of part of the model. Andrew suggested that at first we just require forks of individual subcomponents be made manually.
  • Poul noted that we should have a tracker item for discussing this. Most of the discussion on this has occurred in face-to-face meetings. Tommy is not sure that we have enough ideas to create a tracker item yet. The consensus was that we still want a tracker item on this.
  • Action Item 2: Tommy will check to see if there is already a tracker item on this, and if there isn't, will create one.
  • Tommy has also been working on developing a way for casual web users to browse through models in PMR2.

PCEnv and API updates

  • Andrew has already provided his updates above under the priorities topic.
  • Justin has been working on small bugfixes. He has fixed a regression from the latest Mozilla update, due to the security preferences.

PMR2 Progress update

  • See above under the embedded workspaces agenda item.

CellML Spec version 1.2 progress update

  • Andrew has been working on his talk for the Seattle workshop, which makes some reference to proposed CellML features like data types.

Other items

  • Poul has received some correspondence from David (Andre) Nickerson about the International Conference on Biomedical Engineering (ICBME) in Singapore, which is an conference that has been based in Singapore for the past few sessions, and which focuses mainly on work done in the Asia-Pacific region. Andre noted that we could have a CellML session at this conference if we can get six fee paying speakers to attend. Poul thinks that we should encourage this.
  • Poul has met with Shankar Subramaniam (who heads the Alliance for Cell Signalling) in San Diego. Shankar has expressed an interest in providing input on the CellML repository. Poul thinks that we could be linking into the AfCS databases to get parameters for models, and have models which are updated from the database (in some sense, making the CellML repository part of a bioinformatic pipeline). One approach would be to simply update models manually and have a database link back to the AfCS databases. Another approach would be to automatically update the models, possibly using some sort of language like XSLT to convert AfCS information into models, or by constructing CellML models from an OWL source which can reference AfCS information as well. Shankar has provided a set of notes on the AfCS databases to Poul.