ABI CellML meeting minutes 2009-12-16

Present: Dougal Cowan, James Lawson, Catherine Lloyd, Randall Britten, Tessa Paris, Mark Hanna, Ethan Choi, David "Andre" Nickerson, Tommy Yu, Poul Nielsen, Andrew Miller, Justin Marsh

Apologies: Mike Cooling, Peter Hunter

Last week's action items:

1) Catherine to get in touch with Jon Monk at Uniservices to ask about IP issues.

  • Done.

2) Catherine and Andrew to get in touch with the users to direct them to the clearer descriptions in the 1.2 draft spec.

  • Done.

3) Dougal to update the 1.0 spec to fix the broken images.

  • Done. These images are now copies of the JPGs from the 1.1 spec. Dougal will update all these images to new PNGs in the near future.

This week's agenda:

1) (Catherine) Terms of use for CellML models and associated content hosted on models.cellml.org. Tracker item 2056. Suggestion that we use Creative Commons Atrribution 3.0 http://creativecommons.org/licenses/by/3.0/ together with a specific statement about attribution.

  • Catherine and James looked at the creative commons licenses, and the attribution license looks fairly appropriate. We would need to change our language license to this as well.
  • Jon Monk said that we should have a disclaimer, similar to Biomodels.
  • There was a discussion about attribution vs. citation.
  • Andre asked if everyone submitting models to PMR2 will be required to do so under the CC license - this is what Peter is in favour of.
  • Poul said that he couldn't see a problem with people applying other licenses to their own work in PMR2.
  • Randall said we will need a good way to indicate which license applies to each model - possibly in metadata.
  • Poul said that we will need a way to indicate the licensing for images, and this will have to be obvious for images found by Google™ searches, for example.
  • Andrew said that we will need to consult on how to word the license so that it is watertight that everyone using PMR2 has agreed to it.
  • Andrew explained that all of the current models have been put in under no license, and we will have to get agreement from all parties to apply the license to their work.

2) (Catherine) Should we/can we update the current CellML 1.1 tutorial:  http://www.cellml.org/getting-started/tutorials/cellml-1.1-tutorial.  At the time this was intentionally vague as we did not have a "best  practice" 1.1 format.  To my knowledge we still don't have any "best  practice" guidelines, but we probably need to address this as we now  have users who are interested in building their models in 1.1 and they  need more instruction.

  • From Catherine: "Currently we have Andre, Mike and Jonna frequently using 1.1.  I have  used it a bit, and Sarala has a method outlined in her modularisation  paper.  I believe all of us approach it slightly differently.  Are we  able to come up with a single best practice set of guidelines?"
  • Poul said that we need a group of 1.1 users to get together to try to come to a consensus.
  • Poul said that Sarala and Matt should be involved as well.
  • It was agreed to have a meeting after January 17th to discuss this.

3) (Catherine) Idea: Should we have a CellML 1.1 model library building effort early next year (2-3 days set aside)?  We'd have to agree on a common format (template?) before hand.  We have been talking about a component library for a while now and the timing might be right as we currently have a larger-than-normal team of curators to work on this.

  • Covered under item 2.

4) (James) Additional features in Javascripted SVG sessions.

  • James reported that Mark has been working on the Javascript, and has improved the speed. Mark has also added features: highlighting of SVG diagram elements on mouse-over, and blinking of graph traces on mouse over of the relevant SVG elements.
  • James asked what else we could do with Javascript sessions.
  • Catherine asked if we could have zoom & pan for the SVG diagrams.
  • Randall asked if it was possible to do away with the copy & paste of Javascript into the XUL file of sessions.
  • There was an extensive discussion about the session Javascript becoming a library.
  • It was agreed that this issue requires further discussion.
  • Mark wants to add highlights that indicate whether an SVG element's graph trace is on or off.

5) CellML core specs - update
   1. General

  • No update.

   2. (Andre) Completion criteria (e.g. for reset rules/events): comparison with SBML completion criteria.

  • There was a discussion about how we should be deciding on specification completion, with comparison to the methods SBML uses.
  • Andre said that we should be moving away from specs that have no support [from applications].
  • Randall suggested that the completion criteria could be "the feature works in the API".
  • There was an extensive discussion of specifications and APIs.

6) Should the preferences improvements (currently on the improved_prefs_dec09 branch) be included in the upcoming OpenCell release?

  • There was a discussion about how reliable the preferences are, and how they compare to those found in the PCEnv 0.6 release.

7) Metadata specifications (general, citations, graphing, simulation, other)

  • No update.

8) cellml.org update

  • James has replaced the broken tools index.
  • Randall asked if there is a link to the draft spec yet; this has not yet been added. Randall suggested adding a link to Andrew's draft spec on github.

Action item 1) The cellml.org team to add a link to the draft spec to the site.

  • There was a discussion about how to get a readable version of the draft.
  • Dougal has fixed the 1.0 spec images.

9) OpenCell development update

  • Justin has tested recent snapshots on Julian's Mac, and they are working. They do not work on Catherine's Mac.
  • Randall asked whether anyone has tested the COR-like view.

Action item 2) Justin to find out which is the last snapshot with COR-view enabled, and mail to team@cellml.org, so that people can try it, e.g. Mark, Ethan, Tessa, James, Catherine.

10) CellML API - update

  • Randall reported that Geoff is trying to plug Kinsol into MATLAB.
  • Andrew found that there is no way for CVODE to integrate reset rules.
  • Andre said that there has been a discussion on the Sundials page about this issue.

11) PMR2 development - update

  • Tommy has hosted the release candidate, which is undergoing testing.
  • Tommy has fixed some rendering bugs.
  • Andre has tested embedded workspaces; these don't seem to make much sense under the current single level hierarchy.
  • There was an extensive discussion about embedded workspaces, hierarchies, and unique IDs.
  • Andre suggested per-user folders under which all of that user's workspaces will appear.
  • Justin suggested github as an example of how to do this, the username pre-fixes "workspaces"
  • Andrew said that PMR2 0.2 could be released as is, but then have a new scheme in 0.3 - all of the old workspaces will then need to have their old URLs supported through redirects or similar.
  • There was a discussion about migration to a new hierarchical system.
  • Based on the above decision to implement workspace organisation under each "username" folder, PMR2 0.2 will be deployed as is, subject to completion of release procedure.
  • Randall asked when PMR2 0.2 will be ready to deploy; Tommy said it should be deployed this Friday.

12) Repository contributions update

  • Mark has been working on Javascript sessions, and has not done any model work.
  • Ethan has finished the Leloup Goldbeter 1998 model.
  • Catherine said that she picked two categories of models for the summer students: Goldbeter models because the papers are very accurate, and Bertram models because the original code is available.
  • Tessa has curated the Bertram 1995 model, and has created a CellML Bertram 1997 model which should be pushed by this afternoon. The 1997 model is an updated version of the 1995 model.
  • Randall asked if creating this updated model showed any opportunities to use the modular aspects of CellML 1.1. Catherine said that it could be done, but for now 1.0 is still easier.
  • Geoff has created another large Dan Beard model.
  • Randall mentioned that Dan Beard's group has a system to generate MATLAB code from more structured information about the system they are modelling, perhaps Geoff would be interested in looking at a CellML equivalent.
  • Andrew reminded everyone that the Haskell based approach that he proposed for CellML would allow for such structured information to be directly represented, i.e. CellML models would describe the key system structure and a method to transform this to a computable form (e.g. lower level mathematics). Andrew is working on this now as part of his PhD.

13) Sign up for the workshop!

  • Catherine encouraged everyone to sign up for next year's CellML workshop.