Thread: [SED-ML-discuss] Model identifiers

SourceForge Headquarters 225 Broadway Suite 1600 San Diego, CA 92101 +1 (858) 422-6466

Hi all,

In several places in the SED-ML L1V2 specification, notably section
2.2.2.1, the use of MIRIAM URN's (or URI's) is recommended for
identifying models. BioModels DB identifiers are used as an example of
this "best practice". But as far as I know, following discussion at
COMBINE last year, BioModels DB identifiers are not persistent
identifiers for the same resource - inasmuch as a particular model
(SBML document) identified by a given ID can change between releases
of the BioModels DB.

If my understanding of this is correct (and it could well be very much
the opposite of correct), then BioModels identifiers suffer the same
problem as models in the Physiome Repository in that you need to know
not only what model to use, but what version of the model to use, in
order to unambiguously identify the model. For models in the Physiome
Repository this would correspond to the actual version in the history
of the model and for BioModels DB it would correspond to the database
release number, I guess.

Now, the Physiome Repository provides URLs for all versions of a given
model or you can keep the SED-ML document in the same workspace and
use relative URLs to keep everything at the same version. But we have
not yet been able to come up with a useful scheme to add this
repository as a MIRIAM resource.

Does this mean that there are no useful MIRIAM resources which provide
model identifiers? Should we look at removing that recommendation from
future SED-ML specifications? or, better to look at how we can define
MIRIAM resources which do meet the requirements of the first paragraph
in section 2.2.2.1 of the SED-ML L1V2 specification? Something to
discuss at HARMONY perhaps...

Cheers,
David.

Thread: [SED-ML-discuss] Model identifiers

SED-ML

sed-ml-discuss