Talk:Registry of Standard Biological Models

This is just an idea that has germinating very very slowly with me for a while. Its really just the next logical extension to the parts registry rather than a different idea. Any feedback welcome.--BC 22:28, 22 Sep 2005 (EDT)

Follow up on registry of biobricks models

I fully support the idea of building such a repository of biobrick models. As you mention, it is complementary to the physical DNA repository. Storing, sharing and improving models would be key to be able to move to a Synthetic Biology CAD approach (it seems to be the sense of history in any other engineering field: finite element modelling, electronic systems, control systems..).

However, you describe models having already in mind a way to solve them (Continuous approach with ODEs), it could be important for the registry to simply describe the system to be considered without any constrains on the way to extract their properties. No definitive modelling approach has proven to be more relevant than the others, it should always be possible to explore a given model with different strategies and maybe different purposes. Sometimes, you could only be interested in exploring topological features of the models (patterns, motifs, interlocks) . When it comes to dynamic modelling, you may want to use a stochastic, continuous or hybrid method. The description of the model should be open enough.

As you mention also, an other key feature of the registry would be to provide modular parts. So that you can build a more complex model in a drag and drop fashion. it should also allow the ability to build multiscale models. I would be interested to explore existing standards such as SBML or CellML in order to see if they could be used as a description language for Standard Biological Models. SBML has a strong community behing it and many compatible software to design or integrate models. CellML has developed a nice modular approach. -- User:Vincent - 13th june 2006 -

• BC 15:55, 29 August 2006 (EDT): Hey Vincent, I think you are right, it would be good to make available a generic description of the model (species, reactions, rates) without enforcing a solution method. SBML/cellML might well be good candidates to specify device models in a general way that would still allow users to build models in a hierarchical fashion.

There would need to be standards set such that it is easy to compose models of multiple devices possibly contributed by different individuals. Of course this is the exact same issue that the physical registry is trying to solve:) This was what I was trying to fix by picking a popular solution method (ODEs) but mabye some of the markup languages would be useful to retain solution method flexibility while still permitting a modular approach to model composition.