Meeting Minutes 03 October 2007

Present: Randall Britten, Peter Hunter, James Lawson, Catherine Lloyd, Justin Marsh, Andrew Miller, Poul Nielsen, Tommy Yu

Review of last week's action items:

2007-09-26-1: Tommy sorted out timeframe to upgrade to Plone 3 with Matt, but still need a tracker item be created for this with hours estimation

2007-09-26-2: Catherine went through the site to find broken links

• Not many broken links found, most instances are within the documentation of models within the model repository.

• Other broken links are fixed.

• Roadmap could use some work

  • Could probably be better handled by tracker, have it generate by tracker but it could also be assembled by hand with the information contained within.

  • Poul suggested that it should describe medium to long term details

  • It should also be reviewed from time to time

• Publication page is out of date, last reference was from 2004.

• CellML related efforts should probably contain more links to other projects, to signify that this is a global effort.

2007-09-26-3: Randall had sent a mail about redirecting old repository to new one on the cellml-discussion list, and no objection has been heard.

• Action Item #1: Tommy will go ahead and build the redirection page.

2007-09-26-4: Poul has got updates from Matt regarding the requirements document. The state of the document is still not too clear, but it is coming along. Matt also had discussions with Tommy about getting the state of things to where they should be.

Agenda - updates from everyone:

Poul

• Reaction element in CellML model document is being worked on and revised

  • It could use some more examples to assist model authors to migrate off this deprecated feature, as a number of them expressed concerns about the removal of this element type.

  • Action Item #2: James will need to get update from Sarala and Matt about the status of their recent work on biological metadata systems, as we need to develop a formal metadata framework to store the information about biochemistry that is currently described using the attributes of the reaction element.

• Correspondence with Catherine about issues with the Saucerman-McCulloch model.

  • There was a discussion about how to deal with systems of simultaneous algebraic equations in models. The Saucerman-McCulloch model includes a non-linear system of simultaneous equations, and Catherine has been able to express this in valid CellML, but PCEnv and COR cannot integrate such models. Peter noted that he had been in correspondence with Jeff Saucerman, and that he is working on a model without simultaneous equations as a result of correspondence with Catherine. There was some discussion of the computational costs of identifying systems of non-linear equations and also for performing multivariate root finding to solve such systems. Poul thinks we should aim for a generic implementation which doesn't need any metadata to support it, and add to this if this creates problems for a significant number of models.

  • Action Item #3: Andrew will need a tracker item about enabling multi-variable root solver in PCEnv.

  • Action Item #4: Poul should get in touch with the Formulator group at Hermitech

Tommy

• worked on the CMISS site and started on getting the CCGS separation from model repository going.

Andrew

• Cleaned up code and worked on the validator API

• Added support (via a new interface which is optional for implementors to support) for retrieving the DOM nodes corresponding to CellML elements. Access to this level of information is required to support a validation.

• There was a discussion on correctness of CellML models and specification conforming models.

• James noted that one thing he liked about PCEnv at the moment was that it would allow models which were not perfect (including models with invalid units) to run, which is important as an intermediate step when working on models. Andrew noted that at the moment he is working on a back-end service for validating models, but eventually this will be used by PCEnv to provide validation services to the user. At this point we will be able to design a user interface which addresses concerns such as these. Poul was concerned with the possibility that we would be deciding to 'accept' some models which are invalid CellML. Andrew noted that units problems in the mathematics should be a warning and not a fatal error, and that we should design the user interface to make it absolute clear that models which are invalid CellML should not be distributed as CellML models. James also noted that we could use the validation service to check that models are valid before they are accepted into the repository.

• Andrew noted that nothing seems to have happened in terms of getting a multi-developer continuous build system to replace the existing single-developer system set up on Andrew's computer. Andrew thinks that this should be a high priority now that Justin has started. Randall hasn't heard anything back about this from IT yet.

• The status of the tracker is still the same, but the date to go public is still set at 12 October.

• Tracker was rebranded with the Physiome skin, but still need to retain reference to CellML (via logo, perhaps)