Ontology Sub-Group in the GCP Information Systems Platform and Network Implementation Workshop

Discussants:

Masaru Takeya, Gloria Despacio-Reyes, Serge Gregorio, Alexander Garcia-Castro and Richard Bruskiewich

Strategy of ontology work in the GCP information system has been discussed by ontology sub-group in the Schema design, Middleware and Web services group

One practical way to start ontology work under the GCP is to integrate GCP ontology efforts with the existing ontologies.  We have discussed that to develop ontologies for the data models, first we have to work around the existing ontologies so far, i.e., GO, PO and TO.  The way to do this is to download these ontologies and query each data model so that we get what is present and not present.  Then we will know what particular terms are not in the present ontologies.  The GCP can then work on the terms that are not in any of the present ontologies.    

We have also discussed that in the development of PO, there is a need to find more biological relationships as the present 3 relationships being used in the PO (is_a, part_of, develops_from) are quite limited to carry or to represent the diversity of plant anatomy and structures and development.  One suggestion is to use the biological relationships used by Doi et al in building the rice ontology.

Serge presented the GCP OntologyView, a web application that allows users to manage and browse various ontologies (e.g. GO, SO).  The ontologies are stored in the GCP database, a MySQL database that adheres to the Chado schema.  At the time of the presentation, the GO and SO ontologies were “pre-loaded” into the database by Jun. In a later discussion, Gloria requested a feature that allows a user to search for a term across several ontologies.  This is currently under development by Serge. 

It could be that once a term is not found in any of the existing ontologies, it can then be added into a new ontology, tagging it now with a new GCP identifier or accession.  In this way, we get no duplication with presently existing ontologies.  Perhaps, users can also “borrow” terms from other ontologies, revising them, and adding them to the GCP ontology.

The 3-dimentional browser called OntoRama was introduced by Alex Garcia and if it could be used to develop GCP OntologyView.

Richard comments we should designate new GCP cv & terms with a GCP dbxref accession identifier, not a cv_id. GCP terms could be tentatively assigned to GO/PO/SO/TO, though contributing them to those ontologies would be logical at some point. With respect to Protégé, we need to identify a suitable schema representation, or mechanism to convert Protégé I/O to Chado/ICIS. In the 3-dimensional browser, it would be better to steer clear of Xerox patent of hypertree.