XTM vs RDF/OWL?

Background on the above issue, plus links to others then pursuing the same conversions appear in this 8/15/05 writeup (copy in /patterns) which a few of us in BioPAX tried to help Nada submit (unsuccessfully) for ad hoc funding by Google.

[Update of 6/30/08: This conversion (equivalence?) problem has long been significant for Modeler.  It needs to output EITHER of the above formats, in the UMLS ontologies of choice.  It can, by using appropriate screen scrapers - or go to CTM or Common Logic, or IKL, or CGIF or whatever   We plan to keep Lextender a format-agnostic doc tool; add Turtle support in it for SW semantic net menory systems; focus new work for both BioPax and NLU on portable associations [4] and PSAIDs [6], and wrap all Java APIs in Lisp or WORDS predicates.]

[1] http://www.nadahashmi.com/SummerOfCode.htm 

This spinoff proposal remains on the BioPAX wiki (copy in /patterns).  It expresses an approach to releasing formal OWL-DL ontologies via Modeler, which would better exploit modularity and require better documentation

[2] http://biopaxwiki.org/cgi-bin/moin.cgi/ReleaseToolsProposal

Independently, it also spec'd Barry Smith's conventions on modeling physical things, which at the time seemed a good idea because they linked BioPAX to OBO...

[3] http://biopaxwiki.org/cgi-bin/moin.cgi/CorePhysicalOntologyGuide

An important OWL-DL assoc-topic design pattern on associations and topics was spec'd (and later tested) that could (almost) formally re-express these key XTM objects as restricted classes.  Note: [4] does not allow Occurrences to have scope.  To fix, define them as a blank node with DataProperty String value(s) and ObjectProperty scoping topic(s).. 

[4] http://www.lexikos.com/exhibit/cms/biopax/assoc-topic.ttl

A second worked example below shows how this ontology works for a complex "attachment" model, which was at the time driving people crazy with alternate models of the same thing.  . 
.
[5] http://www.lexikos.com/exhibit/cms/biopax/attachments.ttl

Better documentation and added care on physical/abstract conflation were to be enforced by requiring PSAIDs as part of the registration process for terms in each ontology module.  That could be conveniently asembled in Modeler from various sources.

[6] http://biopaxwiki.org/cgi-bin/moin.cgi/Self-Annotating_Identifers

A bulk data import service from existing data bases that could implicitly add PSAIDs to children of BioPAX classes, and help validate them, was also suggested as a new web service:

[7] http://www.lexikos.com/exhibit/10questions/interactions.jsp

An alternative import process would involve selectively scraping the screens of a loaded Modeler, with a scraper-embedded translation that used [4] and [6], as further discusssed in....

[8] http://www.lexikos.com/exhibit/cms/biopax/usecases.htm

None of the BioPAX proposals have (yet) borne fruit, as they arose just as a major group schism was erupting over the necessity of consistency in OWL models.  It grew, virtually ended effective group work, and persists to this day.