| Author | Topic |
Posts: 961
Registered: October 2003
|
|
Re: SBML L2v2 specification vote #4: References to controlled vocabularies
|
03 Jan '06 15:39

|
 |
|
bshapiro> Also suggest: If a tool does not understand SBO terms
bshapiro> it should be able to ignore them completely, i.e., not
bshapiro> required to write any of the SBO terms back out. I
bshapiro> would suggest the following:
shoops> No, I disagree I see the sboTerm similar to
shoops> annotation(s) where it is good practise to keep
shoops> them as long as you do not modify the annotated
shoops> object itself.
Stefan, I agree with you that it would be good practice. I
guess my hesitation and the reason I wrote that I thought
they should be stripped, is that it seems like a model could
become inconsistent. An object with sboTerms in one part of
the model, even if left untouched, might conceivably become
inconsistent by a change made elsewhere. HOWEVER, I don't
know if that's ever going to be the case. What's needed
here is to work out some detailed examples and see if
there's a way this situation could arise. If not, then I
would agree with you about striving for keeping the sboterms
as much as possible, rather than removing them.
MH
|
|
|