Detecting Inconsistencies in the Gene Ontology Using Ontology Databases with Not-gadgets

Paea LePendu¹, Dejing Dou¹, and Doug Howe²

¹Computer and Information Science, University of Oregon, Eugene OR 97403, USA

²Zebrafish Information Network, University of Oregon, Eugene OR 97403, USA

Abstract. We present ontology databases with not-gadgets, a method for detecting inconsistencies in an ontology with large numbers of annotated instances by using triggers and exclusion dependencies in a unique way. What makes this work relevant is the use of the database itself, rather than an external reasoner, to detect logical inconsistencies given large numbers of annotated instances. What distinguishes this work is the use of event-driven triggers together with the introduction of explicit negations. We applied this approach toward the serotonin example, an open problem in biomedical informatics which aims to use annotations to help identify inconsistencies in the Gene Ontology. We discovered 75 inconsistencies that have important implications in biology, which include: (1) methods for refining transfer rules used for inferring electronic annotations, and (2) highlighting possible biological differences across species worth investigating.

LNCS 5871, p. 948 ff.

Full article in PDF | BibTeX

lncs@springer.com
© Springer-Verlag Berlin Heidelberg 2009