Meeting Minutes 22 August 2007

Present: Randall Britten, Sarala Dissanayake, Alan Garny, Peter Hunter, James Lawson, Catherine Lloyd, Andrew Miller, Tommy Yu (speakerphone).

  • There was a discussion of the process for how models should be reviewed.

    • Consensus at the meeting was that it would be worth sending out new models to a RSS feed and mailing list, but that this shouldn't be the existing mailing lists because not everyone wants to receive so much traffic which would be mostly irrelevant. James will e-mail the cellml-discussion list to request feedback on this.

    • Randall thinks that we should get more feedback from the community to see who would be interested in this and get some more suggestions.

    • The possibility of also including model authors in these notifications was discussed. James and Catherine already do this manually, and James thinks that we are better off continuing to use hand crafted e-mails for this, which also help to promote the CellML project amongst CellML authors.

    • The process around the Auckland meeting and issues discussed there was considered:

      • There was agreement that it was always good to get community input before making final decisions.

      • Peter feels that a formal process like that used by the IESG (the Internet Engineering Steering Group, http://www.ietf.org/iesg) would slow things down and create too much bureaucracy to be applied to simple changes to the website.

      • It was agreed that the process is different depending on what sort of decision we are talking about:

        • The CellML specifications have the most formalised process, with the issues thoroughly discussed on the mailing list, with the Auckland group only making final decisions to resolve controversy or summarise the community consensus.

        • The CellML repository / CellML website is an Auckland developed and run project. However, community input from users is very valuable to assist us to make final decisions and so time should be allowed to let them provide this input.

        • PCEnv is developed in Auckland and will soon have developers from elsewhere involved. Decisions over how Auckland resource is used will always be made in Auckland. Andrew suggested that there are two separate types of decision - how resource is allocated to working on changes to PCEnv, and secondly, whether changes are incorporated into the official version of PCEnv. Andrew thinks that there is always going to be processes local to each developer for the former, but that we need to establish something to deal with the latter to involve other developers such as Alan when he is back in Oxford. Peter thinks that this is something to consider long term when there is more need for this.

    • The process with regards to the 'moderation bit' on the team@cellml.org mailing list was discussed (clearing the moderation bit allows list members to post. At present, anyone can subscribe, but new members have their moderation bit set). It was decided that all decisions to clear the moderation bit should be made at the Auckland CellML meeting.

    • Andrew noted that Taishin Nomura has subscribed to the team@cellml.org mailing list and attempted to post. Andrew approved one of his earlier posts, and he has since tried to post again. Andrew proposed that we clear Taisin's moderation bit. There was no objection.

    • Alan noted that there were a number of other CellML tools out there that we could collaborate with to reduce repeated effort.

      • Taishin's group in Japan, as well as several other groups in Japan.

      • The Continuity plugin developed in San Diego.

      • Andrew noted that we need to develop the PCM-Java bridge if the CellML API is to be useful to the Japanese groups.

      • Peter suggested that we should set up a system allowing for official mirrors of the physiome website / CellML repositories in Europe, the US, Japan as well as the existing South Pacific Auckland site and try to encourage groups to become part of the mirror network instead of starting separate repositories of models.

    • Alan raised the issue of the process for handling meeting minutes.

      • It was decided that the best process would be for Tommy to write up the minutes, notify the people present so they can submit corrections, and a few day later, publish the changes and announce them to the CellML mailing list. Minutes are frozen after this time and cannot be changed.

      • Tommy is not back until some time in late September, so until then the current process will continue as details of exactly how this will be achieved in Plone still need to be worked out.

    • Tommy spent the last week discussing curation flags with Matt (as well as some CMISS related work).

      • James was interested in receiving a copy of the mail.

      • It was agreed that Tommy would send them to team-cellml so everyone can follow them.

    • Sarala has been working on annotating electrophysiological models, as well as developing a set of common rules for annotation which apply across different types of models, and has met some difficulties here.

      • Peter, James, and Sarala agreed to meet afterwards to discuss this.

    • Peter has been involved in discussions of how to deal with the Saucerman-McCulloch model. It can be represented in CellML but solving it requires that systems of simultaneous algebraic equations be solved at each timestep, which no tools currently support. Peter is keen for PCEnv to support this. In discussions with Andrew, Andrew noted that additional metadata may be required to efficiently solve such problems as even the time taken to determine which equations form a system of simultaneous equations to solve grows exponentially in the number of equations, and there may be a need for starting estimates of solutions. Andrew noted that in some cases we may be able to use the component structure of the model to form a heuristic to optimise this, but Alan thinks that we shouldn't rely on modellers to do this.

    • James has been looking at the Saucerman/McCulloch model as well. He now has the Berkeley Madonna model and the MATLAB code. He has also been looking at two MAP kinase models (Bhalla and Iyengar, and Huang and Ferrel), both of which require more work. The Huang and Ferrel model has missing parameters, which he has tried to recover from the MATHEMATICA and MATLAB models. He noted that the MATHEMATICA model was completely different in terms of equations from what was published in the paper - it is unclear if they are mathematically equivalent, but James is looking into getting the equations from MATHEMATICA into CellML. James has also been working on his model wiki.

    • Catherine has been coding up models. She has fixed a number of models she developed some time ago, and has also been working on the Saucerman-McCulloch model and its prerequisite models.

    • Andrew has put a patched version of Mozilla into the Physiome Subversion to make it easier for developers to get PCEnv going, and set up a workflow via git allowing Mozilla updates and our updates to be merged before being put into our Subversion. He has also got an unprivileged Javascript API for manipulating graphs in PCEnv going (which can be used from e.g. SVG diagrams embedded in sessions to create interactive diagrams). He has also changed the layout of the model tree view to add a new 'units' column, and has fixed a number of bugs relating to the handling of bad models that were blocking Jonna's progress. He has also implemented Support Vector Machines embedded in CellML and has found that he can predict knock-down microarray data, when measured by the median of residual squares, better than taking an average over all microarrays.

    • Randall has been adding items discussed in the meeting with Andrew, Alan, and COR users last week into the tracker. There is a follow-up meeting on this today at 2 pm.