ABI Meeting Minutes 2008-06-04

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

Apologies: Edmund Crampin, Peter Hunter


Review of last week's action items


1.)  James has asked Gareth de Walters about getting a folder set up on the ABI CMS for notes / work on the proposed paper about the CellML repository

2.) Mike sent the MSword documents of his introductory CellML tutorials to Catherine who proof-read them and then sent them to Tommy, who converted them to HTML and put them up on cellml.org
  • They currently don't have their own category on the left side-bar of the website, but the group thinks this is a good idea.
3.) The item from 2 weeks ago was added to the agenda for the cellml.org content & design working group, who are currently conducting a content audit of the website and are thus reluctant to add too much new content right now.


Agenda for this week


Geoff - has an example of a model where stochastic modelling is required - good test case for Andrew?
  • Geoff was wondering whether Andrew would be able to make use of a paper he has found that mixes ODE systems with stochastic modelling as a test case for stochastic modelling in CellML.
  • The model describes volumes of distribution of methadone for the major internal organs, and then uses a Monte Carlo simulation to model this process in a population, with each individual being described by the ODE model.
  • This is indeed a useful test case - the more different kinds of stochastic systems we can get test cases for, the better.

Abhishek - progress on representation of time delays in CellML
  • Abhishek is working on modelling of the endocrine system.
  • In endocrine systems, molecules have to travel from one part of the body to another to have their effect; time delays are useful abstractions of this diffusion/transport process, which is realistically too complex to model realistically at present.
  • We are definitely interested in being able to represent time delays in models in CellML, but it is not a short term goal.
  • Because Abhishek really needs to be able to represent time delays in his models, he would be prepared to put some serious time into developing this part of the proposed CellML 1.2 specification.
  • Randall suggested that we could release CellML 1.2 early with time delays added along with a few new features that have already been fairly-well finalised, such as reaction element deprecation.
  • Andrew was concerned about the time-consuming process of drafting and releasing specifications - time would be wasted.
  • It is possible that in the meantime (that is, before CellML 1.2 is released,) description of time-delay systems could be a non-standard addition to CellML 1.1 so that Abhishek can get on with his modelling.
  • These models would have to be flagged properly within the model repository.
  • There was discussion of whether a model of the circulatory system could in fact be leveraged to describe hormone transport. The Guyton models might be useful for this, or the anaesthesiology models which David Cumin described in a recent seminar at the Auckland Bioengineering Institute.

Tommy - making the PMR2 prototype available in order to get feedback
  • Tommy has given the VM of the prototype to James, who has had a brief play with the system.
  • He would like to distribute the prototype to more people, including Mike and Catherine to get their feedback on the software.
  • James has found the modification history system excellent, but echoed concerns that Mike raised in an earlier meeting that masses of modification notes creates a mess - we will need to do something to tidy this up.
  • Randall noted that we might be able to use the 'expose' functionality to address this.
    • Anyone can commit a model, but only curators should be able to 'expose' a model.
  • James asked Randall if there were any specific questions we should be asking when we are testing the software.
    • Primarily, we need to know if the prototype handles CellML 1.1 models effectively, as that is why the new repository was originally commissioned.
    • We need to make sure the repository handles completed 1.1 models, but what is more important is testing out the functionality for collaborating on modular 1.1 models.
    • The current VM-based setup of the prototype doesn't really support collaboration, as users each have a separate instance.
Action item 1.)  Randall would like Tommy to look into hosting a single instance on his machine that can be accessed by James, Catherine and Mike.

Action item 2.)  James, Catherine, Mike and Tommy will meet to test the prototype at 10am on Friday.*
* - meeting time changed to Monday 09/06/08

PCEnv/CellML API progress update
  • Justin plans to freeze PCEnv for the 0.4 release on Monday 09/06/08
  • Catherine wanted to know when the DAE solver will be available for her to keep working on the Saucerman & McCulloch model?
    • Andrew said it is currently accessible by the command-line.
    • It won't be available in PCEnv 0.4, but should be available in a post-0.4 snapshot relatively soon.
  • Andrew has been able to obtain the numerical solution for his generated code using the Levenberg-Marquardt algorithm.

Video of 2008 CellML Workshop
  • James has been working on the video he took of the 2008 CellML Workshop and the Waiheke Multiscale Heart Modelling conference.
  • We have a massive amount of data (tens of gigabytes) - the video needs to be compressed without compounding some of the fuzziness in the video and audio feeds.
  • We could possibly use Google Video to host this information, but we would have to double-check copyright issues and check with the presenters that this is okay.
  • James mentioned that for next year's CellML Workshop, we could use a seminar room in the new University of Auckland Business School building, which are apparently all wired up with video and audio recording hardware (this needs to be double-checked.)

Editing RDF metadata
  • Catherine is concerned that RDF that she hand-codes into CellML models in human-readable format is congregated at the start or end of the file chewed up and spat out with alphanumeric identifiers in a format not readable by humans when the models are fed through PCEnv, COR and the repository.
  • Because this information is then non-readable by humans, there is no way to edit it.
  • Therefore there is a need for RDF metadata editing capabilities beyond those currently provided by the repository (allows editing of authorship, modification histories etc.) and PCEnv (allows editing of simulation metadata,) so that information such as ontological annotation can be recorded and edited.
  • Randall suggested that people who are interested in editing RDF metadata could check out some dedicated software. He has mentioned XML Spy before, and although it is not open source or free (a 30 day trial version is available,) he rates it highly.
    • We should search for some open source alternatives.
  • Metadata editing capabilities in PCEnv and the repository need to be extended - we can continue with specific instances of metadata types, but also we also need some kind of free-form editing capability. If we can find an open-source RDF editor, we may be able to leverage it.

Clashes of RDF metadata produced by different tools
  • James noted that the simulation metadata created by PCEnv often clashes with the processing of metadata by the repository
  • When he changes simulation metadata in a file, uploads it to the repository, then re-downloads it, that metadata can not be picked up.
  • The only solution to this is editing the RDF - he knows how to do this now but to a layman, the solution would be rather esoteric.
  • Thence followed a rather technical breakaway session to discuss the origins of and possible solutions to this issue, which knowledgeable parties consider to be major.