ABI Meeting minutes 2009-01-07

Present: Randall Britten, Dougal Cowan, Andrew Miller, Catherine Lloyd, Justin Marsh, Poul Nielsen, James Lawson, Adam Reeve, Steve Niederer, Jeelean Lim
Apologies: Tommy Yu

This week's agenda:

1) Cell cycle models & SBML events - how can we represent these in CellML?
  • Catherine outlined the issue of repeating models (particularly cell cycle models) and the SBML solution of events.  There is a tracker item discussing this.
  • Andrew stated that we do not want a procedural aspect to CellML
  • Poul thinks we need to brainstorm ways to solve this problem - he is not convinced events are the best way.
  • Randall suggested that the discussion continue on the tracker item.
  • Steve suggested the use of boundary conditions.
Action item 1.) Everyone to think about a solution to the "events" problem.
  • Poul suggested the use of switches, possible in CellML.
Action item 2.) Poul to forward the paper on motifs to interested parties.

2) CellML summer school tutorial - the writing is underway - can we fix a date for this?
  • Catherine said that the writing on this (by herself, James and Mike) is almost done.
  • Catherine asked when this is required by; Poul thinks the first week of March.

3) CellML Workshop - please register if you haven't already done so.
  • Catherine again asked that everyone register.
  • Catherine will do one more push for registrations.
  • We need to put together a programme for our day.

4) CellML API and DAEs, Saucerman, IDA, initial_value attribute etc.
  • Several issues relating to improving the support for DAEs in CellML and the CellML-API have been discussed at a few of the recent meetings and break-away meetings.
  • Randall had a discussion with Steve yesterday, and wanted Steve to mention his views at the meeting.

  • Steve said there are two issues which he would like changed:
      • Code differences - very hard to debug!
      • Random number generator in the method.
  • Both of these have already been reported and discussed before (Code differences - tracker item 1475; extra random guesses - tracker item 1497)

  • Randall mentioned that a way may have been found around the problem, experimental code to do with the 99 random guesses.
  • Poul asked why we need the 99 random guesses
  • Andrew explained the need for the solver to have the 99 random guesses.
  • Steve also prefers the use of initial values for algebraic state variables of DAEs, which is one of the approaches suggested in previous discussions. He also thinks that the random number generator should be removed.  He stated that he thinks the user should be able to choose which trajectory the model follows, so that the user can match their experimental results from the range of possible trajectories.  There was an extended discussion around the problems of both the random guess and initial value approaches.

  • Randall asked if people test how sensitive their models are to initial conditions.
  • Randall agrees that picking initial values is the most practical method, but that it is useful to know when a model lacks information.
  • Poul said that he doesn't think the random guesses approach is the way to go.
  • The discussion was halted so that people can think about the problem and discuss it further at next week's meeting.

5) Repository contributions - update
  • Catherine announced that the Pasek 2008 model is working in PCEnv and COR.  She is waiting for Martin Fink (?) to provide the MATLAB code for the Pasek 2006 model.
  • Jeelean has been working on cell cycle models.
  • Diagrams are being created by the summer students.
  • Poul expressed concern about whether PMR2 will be ready to deploy by the workshop.
  • Randall said that Tommy was starting again in February, but that it is still unclear whether PMR2 will be ready by the workshop.

6) PCEnv development - update
  • Justin has been fixing bugs, and has been making cosmetic improvements to the Mac OS version.
  • Catherine has run PCEnv on the Mac and has been unable to break it; and there was much rejoicing.
  • Steve suggested features for PCEnv which he has been working on implementations of:
      • A tool to take CellML, define variable input and output values, and provide a small set of equations that provide those outputs.  This can be done batchwise to provide large sets of data (data books) for model comparison, comparing numbers of plots.
      • A batch processing sensitivity analysis tool that allows users to perturb chosen parameters by defined amounts.
  • Steve also saw the need for a way to rapidly calculate the limit cycle, to more conveniently find the steady state.
  • Steve said he thinks that these kind of powerful new features are needed to provide the incentive to switch to PCEnv.
Action item 3.) Randall to create tracker items for Steve's ideas.
  • Adam has got the MATLAB code export working with DAEs
  • Catherine wants MATLAB to CellML conversion.
  • Randall wondered if having only CellML to MATLAB might be useful to encourage people to initially code in CellML.
  • James asked if it was planned to export in other languages; Randall said it is.

7) CellML API - update
  • Andrew has added some missing functions to the Java wrapper.
  • The Java wrapper needs testing by somebody.
  • We could do with some documentation on the Java wrapper.
  • Randall said that people need reminding that the Java wrapper is now available for use.
Action item 4.) Andrew to add a comment to the Java wrapper tracker item to ask for feedback.

8) Website redesign - update
  • Catherine reported that the tutorials are largely written, and will soon be up on the development site for people to have a look at.
  • Catherine asked if everyone could please look at current sections of the website (eg. tools section) that they are responsible for, to see if any updates or changes are required.

10) SBML to CellML converter prototype - update
  • Andrew continues to work on the prototype, but it is not yet ready for use.
  • Catherine asked if we have been in touch with Nicolas Rodriguez; Andrew said we have not.