Meeting Minutes 20 June 2007

Meeting Minutes 20 June 2007
Present: Randall Britten, James Lawson, Tommy Yu, Poul Nielsen, Peter Hunter, Catherine Lloyd, Andrew Miller

Randall noted that a lot of the website is now out of date, and he suggested that there be a separate meeting to deal with the structure of the website.

Tommy created a UML class diagram of a metadata schema to store RDF data. It has tables called Person, Annotations, Journal Article, Annotations, Modification Note, Creation Note etc... and could be stored in a relational database. Tommy envisages there being three key components: A model store (e.g. Subversion), a metadata store (in a relational database or ZODB if we decide that we still want to keep it), and a front end.

Andrew asked about the implications of this for the versioning of metadata. Andrew thinks that if the metadata changes this should be a new version, although Tommy was originally planning on having the metadata not being versioned in the same way as the data. This probably needs more discussion.

Randall has been through the roadmap documents he could find on the CellML site because Peter has suggested we might be able to get a new developer to work on PCEnv together with Andrew. It is somewhat out of date, so he has asked for suggestions from within the group. Peter thinks that the philosophy behind PCEnv is for it to be a tool which can be used within the institute, but which is open source and is available for people outside the institute to use. He therefore thinks that any decisions on the priorities of features should be decided on the basis of the usage within the institute (although ideas on features can come through the bug-tracker from people outside of the institute).

James suggested that PCEnv have a more compact way of viewing imports and connections.
Andrew suggested that connections (and mathematics) could be summarised to take James' concerns into account by putting read-only text describing component_1 and component_2 in the value field before component is expanded (which can be edited by expanding the component).
Peter is keen for PCEnv to interact with the repository at some point.
Poul would like to see a feature which makes it easier to break CellML 1.0 models up into parts to form a CellML 1.1 model.
Peter and James think that we need some more domain specific editing support for model types such as metabolic models, gene regulation, signal transduction. Andrew suggested that we might need to develop metadata specifications so we can represent information these communities need. Poul noted that Sarala will be developing something like this as part of her PhD.

James noted that there had been some discussions on the mailing list...
  • Andre has asked about how to update the bioeng publications page. Consensus was he should be referred to Gareth.
  • There was some discussion about the list. There has been no mail on the list, so the group decided to close down the mailing list. To address Matt's concern that people might want to ask questions without broadcasting them so widely, the page about people involved in CellML needs to be updated.
  • The front page still links to the old repository. Peter thinks that the link should be maintained, but it should be moved somewhere much less prominent.
  • The text announcing upcoming features can also be moved to the navigation area on the right hand side, although it should still be made clear that the repository is still under development.
  • Catherine noted that the Pole-Zero law had been put into the repository but the metadata for Mooney-Rivlin had been put on the model, causing the Pole-Zero law to have the wrong name. This has now been fixed.

James has been working on setting up sessions. He could create 'reference descriptions' involving multiple models with parameter tweaks for some models, but he thinks that it is better to fix the models which are in worse state first, rather than making a few very good models. His goal for now is to bring all models up to 2 star curation level.

James also suggested an improvement to how the sort function worked, because different versions are being sorted independently now making it hard to find the right variant. This particularly affects sorting by curation level.

Peter is looking at setting up an electronic CellML journal which would allow peer-reviewed models to be cited in other publications. The idea is that people would be able to cite both the CellML model article and the publication article, therefore giving them an incentive to submit a working model. He has been in contact with Oxford University Press to see if they can help with some of the details.