ABI Meeting Minutes 2008-06-11

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

Apologies: Poul Nielsen, Peter Hunter, Edmund Crampin

Review of last week's action items

1.)  Tommy has set up hosting of the PMR2 prototype on his computer

2.)  Tommy, James and Mike had a meeting regarding the PMR2 prototype but Catherine was unable to attend

Agenda for this week

Mike - Tracker item 31 "Allow integrator points reported to fall exactly on boundaries, for comparison purposes'

  • Tracker item 31 is has priority attribute 'moderate.'
  • Andrew noted that there are many 'moderate' priority tracker items that need to be done.
  • Randall is keen for the developers to concentrate on features / bugs that users feel are important to them.
  • Mike's request is for the API to allow specification of fixed step sizes in integrations. Currently the API uses variable step size, so it is usually only possible to get a value for the time point for 9.999978349 seconds, for example, instead of 10 seconds. This means that proper comparisons between simulations can not be made, and values at round-number time points can not be accurately recorded in publications.
  • Andrew said that it would be perfectly possible to add this feature so that the user can select whether they want to do a fixed time step integration or not, but the project needs to be a priority or it won't get done for a while.
  • Mike noted that fixed time stepping would be inaccurate and it would be better to have the integrator approach time-unit boundaries with variable time stepping, and report output at time-unit boundaries.
  • Andrew agreed and mentioned implementation via a range of smaller integrator runs, one up to each time-unit boundary.
  • Mike noted that the Newcastle (iGEM2008) group will be  using CellML and associated tools in the near future and this feature would certainly help them.
  • Randall mentioned that Andrew, himself and Justin will be meeting with Alan Garny soon, and that he will raise this issue.
Action item 1.)  Randall to get back to Mike regarding time-unit boundary integrator output points.
  • James asked whether voting in Bugzilla might have been useful in this scenario - the priority of the issue could have been left as it was set by the developers, but Mike could have left an indication that he was interested in seeing this feature implemented.
  • Randall and Andrew think that due to the current small project size, direct communication with the developers is probably better.

James - videos of CellML Workshop 2008, video camera broken
  • James has managed to offload most of the video tapes of the Waiheke MSMH conference (not related to CellML,) but the video camera has stopped working. It is no longer able to communicate with the computer for some reason. He has tried to fix it but is stuck, and unsure how to continue.

PMR2 progress update
  • User testing of the PMR2 prototype has begun.
  • Catherine needs to meet up with Tommy and get taken through the software.
  • Mike suggested that some kind of documentation or mini-tutorial explaining the main terminology the site uses and how to navigate it would be useful.
  • James has uploaded some work, including a session file that uses relative links and contains a Javascript-embedded SVG diagram in a XUL container.
  • There is an issue with PCEnv representing the URI of the XUL file as a literal rather than a URI.
  • Mike noted that the 'expose' function exposes specific versions, but that some kind of upper layer is needed that will contain all exposed versions.
  • Tommy has been discussing reworking the CellML citation metadata using the MIRIAM format. He has had David Nickerson and Nicholas le Novere's input on this matter. See tracker item 414 for more information. Tommy intends to develop an absolute citation reference which can be used for searches by PMR2.
  • Randall suggested that a search that brings up non-deprecated exposures of versions of models (that is, versions that have some utility in and of themselves,) that have common metadata. In this case that common metadata would denote that these versions of models are based on the same original publication.
  • James's conception of the workflow that should be used is that the user can expose as many models as they want, but interacts with the models primarily through a search interface that is informed by metadata.
  • Related models will be linked both by metadata and the intrinsic structure of the Mercurial repository.
  • Randall cautioned that referencing URLs in a publication that are supposed to bring up a particular search result listing would require long-term stability of these search URLs. We need to be careful in designing what the search result URL might look like to counter this.
  • Justin suggested that we might want to use a system such as TinyURL to represent long URLs in a stable format.

PCEnv / CellML API progress update
  • Justin has released PCEnv 0.4 release candidate 1, he will be sending an email to the CellML discussion list informing the community. There will be a week of community testing before the actual release of PCEnv 0.4 is made.
Action item 2.)  Justin will send an email to CellML Discussion announcing the release of PCEnv 0.4 rc1
  • James noted that he has been keeping up to date with the snapshots and has found no bugs for quite some time.

  • There was a meeting between Andrew, Justin, Randall, Poul and John Butcher regarding the DAE solver project.
  • John said he thought the current approach being taken is sensible.
  • Andrew noted that the results for the simple model that Randall wrote (that has both an analytical and numerical solution) are coming out nicely.
  • The Saucerman & McCulloch model is not yet able to be solved, however.
  • Catherine has curated this model in COR to the point where the only error messages are about circular arguments, which is to be expected. However the model has been changed enough from the original publication that she is not entirely sure whether it is correct or not.
    • Andrew has not tested the latest version of the Saucerman & McCulloch model yet - Catherine will send it to him.
Action item 3.)  Catherine will validate her latest version of the model against the original MATLAB code for the model.

Geoff - Coding up Dan Beard's mitochondrial TCA cycle / oxidative phosphorylation / metabolite transport / electrophysiology model
  • This is a gargantuan model which essentially models the major biochemical and physiological systems of the entire mitochondrion. Geoff has coded up the majority of it, but there are some binding polynomials that he is still working on.
  • It has some PDE-like equations, Geoff is getting the model ready for it to be plugged into CMISS. The PDE-like equations in this model are non-spatial - they describe differentials of species concentrations with respect to each other.
Action item 4.)  Randall will have a look at the MATLAB code for this model and see if he can decipher what kind of equations these are.

CellML 1.2 specification progress update
  • There is a list of proposals for items in the 1.2 specification that have been fully discussed and just need final confirmation from the CellML Specificaiton Panel.

James / Catherine - CellML Model Repository paper
  • Catherine has written a draft mini-paper pitched as a Bioinformatics Application Note describing the CellML Model Repository
  • The format is either 1000 words and a figure or 1300 words.
  • James has read the draft and made some comments / changes, and will add some further discussion of annotation of CellML models.
  • The paper will then be sent to Poul and Peter for their input.
  • This paper is primarily designed to provide a reference for people who want to cite the CellML repository.
  • Catherine and James figure it is better to write a number of smaller papers rather than one or two monolithic papers.
  • The next paper will probably be written when the PMR2 software is being used, and another paper about using flags to curate models is also in the pipeline.