Meeting Minutes 2008-01-30

Present: Randall Britten, Michael Cooling, Edmund Crampin, John Davidson, Sarala Dissanayake, Catherine Lloyd, Justin Marsh, Andrew Miller, Poul Nielsen, Jonna Terkildsen, Anil Wipat, Tommy Yu


Agenda

Mike: BioBricks

  • David Gilbert is interested in using CellML for the models they are developing, as they were pleased with the results of the translation of one of their models to CellML.

    • They are interested in the metadata to represent information that could not be translated into CellML.

    • Sarala pointed out that she needs to know the structure of the RDF graph.

    • Mike said he will annotate some BioBricks model with RDF to see how it works.

    • Mike pointed out that when he uploaded some model with metadata to the model repository, it insists on using its format and not pick up the RDF metadata he already has in the file.

      • Tommy pointed out that it may be the cmeta:id that didn't match, or that the RDF graph that was presented was not what the repository software expects. He also points out that there is a separate metadata specification that may be of use.

      • Randall suggested Mike try XmlSpy's RDF viewer on a 30 day trial.

    • Metadata specification

      • Poul noted that it could use an upgrade in the light of all this

      • Tommy pointed out that the definitions for the graphs are quite ambiguous, thus difficult for software tools to work with it.

      • Poul also pointed out that Andre is working on the metadata specification.

    • Mike would like to have some examples to pass off to David's group. He doesn't want to create ad-hoc examples that may not reflect what the specification may say.

      • Poul noted that Matt, Sarala, Andre and Catherine are all good people to pass ideas to, as they all have extensive experience in working with metadata.

  • Another topic was a tool that would convert SBML to CellML, but it seems that one may not exist, as conversion between the two will be lossy, unless associated metadata was somehow maintained.


Mike: Copyright for the CellML workshop online presentations

  • Mike pointed out that he will be organizing a copyright seminar which he already notified others on the institute mailing list.

    • The copyright of the presentation - as a precaution we don't want to infringe on the rights of the presenters. Catherine sent out email to all of them and most do not mind being on camera.

    • However, Mike pointed out that if someone used material that is covered by some other publisher's copyright, and it was recorded in the presentation there will be liability issues.

    • It may be worthwhile to get people to sign a waiver to assert that their presentation does not contain diagrams or other materials that they do not have permissions from the copyright holders to use.

      • Mike points out that the University requires special waivers be signed, and thinks they should be sent to the presenters, otherwise it isn't worthwhile.

Catherine: CellML Workshop 2008 Update

  • Got all the abstracts and the program ready.

  • Catherine passed the draft to Andre, Peter and Poul, and Andre got back to her with some changes but otherwise it is ready.


PMR2 Update

  • Tommy is working on getting Mercurial working within the product.


CellML 1.2 Specification Update

  • Over the last week, Andrew has been through his draft text of the CellML Specification to identify any discrepancies with the CellML 1.1 specification, and a few minor technical differences were identified and corrected.

    • He is getting the document into a state that is ready for others to contribute more easily.

  • Mike wondered if compatibility has been broken yet.

    • Andrew said not yet, aside from aspects of CellML which are unlikely to be maintained in CellML 1.2, because we haven't made a decision on all of these aspects yet.

  • It was pointed out that it would be nice for more eyes to look at the document to point out errors that may have been in there.

    • Action item: Justin and Randall will review the specification draft.

  • Andrew: Secondary specification

    • Andrew has a number of issues to work on over the next week. One of these would be to add text requiring that CellML models comply with a secondary specification, and state what secondary specifications should say. These secondary specifications would narrow the general, primary CellML specification to specific types of problems such as ODE Initial Value Problems (but not contradict with it), and individual software packages would be able to implement specific sets of secondary specifications.

    • Poul noted that the arguments for this should be put forward on the mailing list to start off discussions, alongside any concrete specification text Andrew comes up with.

    • Andrew will work on the text in the primary specification about the secondary specifications, but will not at this stage put the effort into writing up a secondary specification.


Buffer overflow issue in CellML API

  • Certain Unicode characters cause buffer overflow in CellML API on Linux.

  • The question was whether it is worth the effort to make a minor point release.

  • Alan mailed the mailing list and he would like a security update release.

  • Justin pointed out it will take about two to three days total.

  • Andrew pointed out that this bug fix also fixes another issue where a model didn't load.

  • This is to be decided later in the after meeting, and discussion will happen on team@cellml.org


John Davidson on CMISS

  • Most models in the CellML Model do not work out of the box in CMISS.

  • Andrew noted that CMISS currently uses the code written by David Nickerson (Andre) some time ago, and requires CellML models to be in a very specific form to work correctly with CMISS.

  • Andrew noted that Andre is now using the CellML API implementation. The exact details of whether he has integrated this with CMISS were not known to anyone at the meeting.

  • The older CellML code currently in CMISS also only supports CellML 1.0. It was noted that not having imports (CellML 1.1) makes it harder to separate stimulus protocols from the core of the model

  • Poul wondered how much effort is involved in getting the CellML API into CMISS.

  • Andrew noted that we don't really have anyone working on both CellML and CMISS.

    • To answer the question, CMISS and the CellML API are written in different languages (Fortran and C++), so it would be difficult.

    • However it is possible to use the CellML API to generate Fortran code, which CMISS can import as a module and execute. This would require a Fortran definition file written for the CellML API so it can be used to translate CellML code into Fortran.