Poul kicked things off by distributing a copy of the Road Map.

Rod Smallwood explained that they submitted a grant application for the development of cell models and TissueML - 1 person for 3 years. They are not yet certain if the money will come through, but if it does (by June 1), they intend for the person funded by the grant to coordinate with Team CellML.

Poul drew attention to some of the main points discussed in the Road Map document:

Matt: “One reason we want to move the reaction part of CellML into ontologies is that other groups (such as BioPAX) have better descriptions of the qualitative information regarding biochemical pathways. There is no reason to recreate something that's already been done fairly well. This is where typing information comes into play.”

“The anatomy ontology can be used to find mathematical models and vice versa.”

Dawn: “Are BioPAX models freely available?”

Matt: (explaining) “BioPAX describes the qualitative aspects of a pathway, information about species and reactions. It does not describe the mathematics. There is currently no centralized database of BioPAX models. You might be interested in checking out PATIKA, a global shared repository (can be stored locally or centrally). The PATIKA group is heavily involved in the development of BioPAX.”

“Eventually we expect to make all of the CellML pathway models instances of BioPAX.”

Dawn: “Where does FieldML fit in?”

Matt: “That's a question for Shane, but the plan is to integrate FieldML models with the anatomical ontology. For instance, you could query for the FieldML representation of the tibia.”

Rod: “How do you handle validation?”

Poul: “There's validation at different levels:”

At this point, the discussion went off on a tangent about the role and definition of models sparked by Rod's comment, “The validation of the model answers the question ‘Is this model specification doing what I specified it to do?’ rather than ‘Is this model a faithful representation of reality?’” Though this discussion was interesting, it wasn't entirely relevant, therefore, I didn't try to capture it in my notes.

Steve: “How is CellML going to be managed for change control and bug fixes?”

Poul & Matt: “CellML is at a point where it needs to be released so that community can drive its direction. 1.1 is at a stable point. We have a UML description, DTDs, and XML Schemas available. We just need to kick the process off.”

Matt: “There will be [three] communities that overlap: The tool development community (bug trackers, cvs repository, platform testing), a community for the development of the specifications, [and the user community interested in building and running the models]. We are close to having this all in place. See the roadmap.”

Matt: (addressing Dawn) “Regarding the modelling of physical conditions, we need to get people like you involved.” CellML core (ModelML) + ontological layer focussed on tissue-level.

Me: “In which direction is TissueML headed?”

Dawn: “No one's decided yet.”

Poul: “We'd like to get your feedback on everything we've talked about. See if it will suit your needs.”

Dawn: “I'm not sure if some of the epitheliome processes can be described by a markup language. Describing the structures my model generates would be a good start. The ML would describe a template for the general structure, then we could create specific instances.”

Rod: “How did you develop CellML? How did you know what to put in?”

Poul: “We had a set of requirements aimed at the cellular level. It turns out that the solution we came up with has a more general applicability.”

Dawn: “I've written down the parameters I'd need for the description at the bladder tissue level (components and variables). I will go over these with Matt.”

Matt: “An ontology is a nice place to start to define what you need to describe.”

Rod: “The point of view from which we approach this is important. Structure first or function first?”

Matt: “Use TissueML to describe states and EmbryoML to describe what these states mean? You might be interested to hear what UMLS have been doing. They are bringing together databases to address how we model developmental processes. They define two databases together and sort out conflicts. The Digital Anatomist project (describing structure/function together) sprung from this.”

[Author's aside: my notes on this conversation were extremely slim (as in, key words only) because I'd run out of paper and was notating in the margins. Matt later fleshed out the missing bits in an e-mail as follows.] Rod talked about structure, function, and then transformations, such as those observed over development. The nature of my comments on structure and function were as follows...