Re: SBML L2v2 specification vote #4: References to controlled vocabularies
03 Jan '06 15:30
mhucka> a) If an sboTerm attribute is present on a KineticLaw
mhucka> object, the 'math' field must also contain a formula.
mhucka> A model containing an sboTerm on a KineticLaw lacking a
mhucka> math field would be deemed invalid.
lenov> I think I disagree. Currently a reaction can be
lenov> defined without a kineticLaw. Although the models
lenov> are not simulatable, they are nevertheless useful,
lenov> if only to import and develop kinetics models.
lenov> An sboTerm attached would increase further the
lenov> usefullness of those SBML files.
OK, if I understand your point, you are saying that it
should be acceptable to have kinetic laws with sboTerms but
without math filled-in?
It seems to me this would put us in the situation (I think
it was Ralph Gauges who pointed this out) that it would
cause all software to be required to understand sboTerms,
simply because models containing only sboTerm and no math
would now be considered valid.
lenov> A possible solution would be to put an optional
lenov> sboTerm on the reaction element. This sboTerm would
lenov> just be an
I think allowing sboTerms on Reaction would be a fine idea.
lenov> Verbatim, but for the symbols, that have to be
lenov> those of the model, not the CV.
Hmm. I think if one used user-defined functions, the
substitution problem would be avoided.
I thought I had already posted the whitepaper on this, but
apparently not (or I can't find it now). It must have
gotten wrapped up in the stochastic/discrete vs continuous/
deterministic proposal we were discussing off-line last
year. Argh. Well, let me just claim for now that I think
there are ways of reducing the symbol-correspondence burden.