ABI Meeting Minutes 2008-06-18

Present: Randall Britten, Poul Nielsen, James Lawson, Andrew Miller, Geoff Nunns, Mike Cooling, Tommy Yu, Justin Marsh, Catherine Lloyd

Apologies: Edmund Crampin, Peter Hunter


Review of last week's action items:


1.)  Randall responded to Mike re: his suggestions for the CellML API's integration service. The issue has since been discussed by concerned parties in further detail.


2.)  Justin announced PCEnv 0.4rc1

  • PCEnv needs to be installed with admin rights in Windows XP when there are spaces in the directory name it is being installed into, such as 'Program Files' and 'The University of Auckland'

Action item 1.)  Justin to create a tracker item to discuss implementation of a workaround for this.


3.)  Catherine spent quite a bit of time last week working on curating the Saucerman & McCulloch model

  • She has created a version purely based on the original MATLAB code of the model, but it appears to be overconstrained.

  • Andrew suggested that Catherine add ID's to the equations in the model to aid in detecting where the superfluous constraints are in the model.

  • The next trunk snapshot of PCEnv will be able to use the DAE solver.


4.)  Randall looked into the MATLAB code for Dan Beard's mitochondrial model that Geoff is working on coding up.

  • Tracker item 1105 discusses the model.

  • Geoff has been in contact with Dan Beard.

  • Currently the MATLAB code generated by COR from the CellML model can be run in MATLAB, but the actual CellML model will not compile in COR. Geoff is now comparing the original code to the COR-generated MATLAB code.

  • It is possible that this is due to the value of some state variables dropping below zero. Tracker item 1106 discusses a proposed feature for the CellML API to allow state variables to be constrained to positive values.

  • Catherine has used piecewise equations to achieve this for some state variables, but Randall does not believe this is possible for those that are involved in ODEs.

  • Andrew noted that reset rules might allow a modeler to address this problem.

  • Poul said it would be better to express these constraints within MathML and find a solver that can handle them.

  • Andrew suggested that being able to express constraints on algebraic variables as well as state variables would be useful.

A breakaway meeting to discuss this issue occurred after the CellML meeting.


Agenda for this week


James - Populating Sarala's ontology

  • James met with Sarala earlier this week to discuss how to begin populating her ontology.

  • Sarala gave him a copy of her tool and showed him how to use it to generated OWL representations of CellML models, and from there, CellMLBio representations of CellML-OWL models. She then demonstrated how to populate the CellMLBio ontology using her tool.

  • James has been thinking quite seriously about how to construct the ontology. Creating rules that automatically populate classes with instances is extremely useful but one has to be careful that the rules do not provide clashing instructions.

  • Poul suggested that Matt Halstead is probably a good person to talk to about this.

  • James also noted similarities between using and creating ontologies / constrained vocabularies to annotate CellML models and creating a system of curation flags.

  • Poul mentioned that the CellML-OWL representation of a model can actually be used to translate CellML into other formats, such as SBML. This would be better than previous XSLT-based methods for CellML<->SBML translation.

  • Mike noted that Anil Wipat from Newcastle is very interested in this OWL representation of CellML for the work his group is doing.


PMR2 feedback discussion

  • James is keen to see modification histories available for not just entire workspaces (as now,) but also for individual files.

  • Randall suggested that model authors should provide a summary of the modification history of their workspace to date when they expose it. The exposed model would then be indexed, but not listed as curated. The modification history summary (amongst other requirements,) could then be vetted by curators as part of the curation workflow.

Action item 2.)  Tommy to notify team@cellml.org of how to access his PMR2 prototype.

PCEnv / CellML API progress update

  • Justin will be releasing PCEnv 0.4 and the CellML DOM API 1.4 at 4pm today.

  • The config script needs to be added, but this will not invalidate the release candidate.

  • We also have a code signing certificate for the ABI which we should use to sign the Windows installer. This will be the only difference between the RC and the release binary.

Action item 3.)  Justin to add config script to release.

Action item 4.)  Justin to use the signed version of Windows installer for release.

  • Andrew has gotten the DAE solver working and has checked the code into the API trunk.

  • The numerical soundness of the approach is still under investigation.

  • Andrew has also been working on a way for more expressive variable names to be used by the CellML CIS when generating procedural code. At present, the names are simply indices.

  • This could be implemented in PCEnv but Andrew has concerns about generating code from models that import multiple instances of the same submodel.

  • Randall noted that this issue extends to the representation of variables in the graph variable selection listing.

  • Randall mentioned that the PCEnv / COR developers had a meeting to plan for PCEnv 0.5 - see tracker items 1089 and 1090.

CellML 1.2 specification progress update

  • Andrew has been working on the proposal to make use of lambda calculus in the specification.

  • James was concerned that this might make the use of CellML more complicated, and was keen for it to be hidden.

  • Andrew countered that leveraging lambda calculus may in fact make using CellML more simple.

Action item 5.) Poul to record his stance on the present list of non-controversial items to be added to the CellML specification via the tracker items for these.

Repository curation

  • Randall would like to add a new item to the standing meeting agenda: that is, a updates on new models being added to the model repository. He has also submitted a news item to cellml.org describing the 'new models' RSS feed and is keen for this information to be publicised more widely.

Action item 6.)  Randall to add item to standing agenda.

  • Catherine has been coding up some private models that have not yet been published - she expects the model authors to refer to these models within their publications via a link to the repository.

  • The paper discussing PMR has been submitted as an application note to Bioinformatics. It is approximately 1300 words long and lists Catherine, James, Poul and Peter as the authors.


Singapore conference

Action item 7.)  Catherine will email David Nickerson regarding this.