SGBN-1 Discussion Notes: Resolutions

(Discussions moderated and notes taken by M. Hucka.)

M. Hucka: One of the questions that we were left with from earlier today is whether PPI implies the need for a 3rd type of view. What was the result of discussions?

H. Kitano: The conclusion seems to be that there's a different model in PPI, and PPI needs a view of this model. It's not the same model that underlies entity-relationship and state-transition diagrams. He proposes the following levels of definition:

• model: something in CellML or SBML, a machine-readable form
• view: this is “how to graphically represent a model”, like an entity-relationship (MIM-style) or state-transition (PDN-style) diagram

• layer
• abstraction
• levels
• versions

P. Ghazal: The conclusion seems to be that if you're talking about protein-protein interaction, it falls under category of an ER type of view. If you're talking about a process, it falls under the Kitano type of state-transition view. If you're talking about a hierarchy, it falls under the Edinburgh type of view. Now here's a different question: does SBGN support data without an underlying model?

U. Dogrusoz: There seem to be 2 model-layer-view constructions going on. The ER and ST are 2 views of an underlying conceptual model of something, but they also act as 2 models themselves. Your diagrams are 2 views of those two models.

(There was additional undirected discussion. Unfortunately, I was trying to think through the issues and failed to take adequate notes during the rest of this session. — M. Hucka.)