CL ARRA progress report for 2010
ONTOLOGY ALIGNMENT AND INTEGRATION: we have supported the JAX ontology editors by providing and executing tools to integrate human cell types represented in the FMA (Foundational Model of Anatomy) into CL. These tools identified generic cell types in common between the two ontologies, and used the Obol toolkit (developed at Berkeley) to identify compositional class expressions for specific cell types (e.g. "smooth muscle cell of jejunum"). These compositional classes can then be integrated automatically, using the cross-species gross anatomy ontology UBERON.
ONTOLOGY EDITOR SUPPORT: we have provided ontology editor support to the JAX editors of the ontology, improving the oboedit ontology editing environment in reponse to curator requests. We have also implemented a variant of the MIREOT (Minimal Information Retrieved for External Ontology Terms) strategy which allows scalable integration across multiple ontologies. We have also devised a build environment for the CL, which allows the easy distribution of multiple versions for different purposes, and uses OWL reasoners to classify and perform logical consistency analysis on the ontology automatically.
ONTOLOGY LANGUAGE SUPPORT: we identified key areas in which the ontology format used by the Gene Ontology was insufficient for representing cell types. Rather than move wholesale to the more expressive OWL (Ontology Web Langauge) formalism, which would have entailed abandoning tools and interoperation with model organism databases, we enhanced obo format using a macro mechanism and implemented a prototype macro expansion engine (http://precedings.nature.com/documents/5292/version/1). This allowed us to have the best of both worlds, and to be able to transition between the oboedit environment, and the Protege4 editors environment, and to use OWL reasoners to classify the ontology.
This work also necessitated creation of a formal specification of obo format and its translation to OWL (http://berkeleybop.org/~cjm/obo2owl/obo-syntax.html), which significantly extends on previous efforts made in 2007, before the OWL2 language was finalized.