Identifying gene ontology areas for automated enrichment

Abstract

Biomedical ontologies provide a commonly accepted scheme for the characterization of biological concepts that enable knowledge sharing and integration. Updating and maintaining an ontology requires highly specialized experts and is very time-consuming given the amount of literature that has to be analyzed and the difficulty in reaching consensus. This paper outlines a proposal for the development of automated processes for the enrichment of the Gene Ontology (GO) that will use text mining techniques and ontology alignment techniques to extract new terms and relations. We also identify the areas of GO whose level of detail is too low to answer the community's needs at large. We have found that although GO's content is well suited to the manual annotations, revealing the coordination between GO developers and GO annotators, there are 17 areas that would benefit from enrichment to support electronic annotation efforts. With this work we hope to provide biomedical researchers with an extended version of GO that can be used 'as is' or by GO developers as a starting point to enrich GO. © 2009 Springer Berlin Heidelberg.