Email updates

Keep up to date with the latest news and content from Journal of Biomedical Semantics and BioMed Central.

This article is part of the supplement: Proceedings of the Bio-Ontologies Special Interest Group Meeting 2010

Open Access Proceedings

How orthogonal are the OBO Foundry ontologies?

Amir Ghazvinian*, Natalya F Noy* and Mark A Musen

Author Affiliations

Stanford Center for Biomedical Informatics Research, Stanford University, 251 Campus Drive, Stanford, CA, USA

For all author emails, please log on.

Journal of Biomedical Semantics 2011, 2(Suppl 2):S2  doi:10.1186/2041-1480-2-S2-S2

Published: 17 May 2011

Abstract

Background

Ontologies in biomedicine facilitate information integration, data exchange, search and query of biomedical data, and other critical knowledge-intensive tasks. The OBO Foundry is a collaborative effort to establish a set of principles for ontology development with the eventual goal of creating a set of interoperable reference ontologies in the domain of biomedicine. One of the key requirements to achieve this goal is to ensure that ontology developers reuse term definitions that others have already created rather than create their own definitions, thereby making the ontologies orthogonal.

Methods

We used a simple lexical algorithm to analyze the extent to which the set of OBO Foundry candidate ontologies identified from September 2009 to September 2010 conforms to this vision. Specifically, we analyzed (1) the level of explicit term reuse in this set of ontologies, (2) the level of overlap, where two ontologies define similar terms independently, and (3) how the levels of reuse and overlap changed during the course of this year.

Results

We found that 30% of the ontologies reuse terms from other Foundry candidates and 96% of the candidate ontologies contain terms that overlap with terms from the other ontologies. We found that while term reuse increased among the ontologies between September 2009 and September 2010, the level of overlap among the ontologies remained relatively constant. Additionally, we analyzed the six ontologies announced as OBO Foundry members on March 5, 2010, and identified that the level of overlap was extremely low, but, notably, so was the level of term reuse.

Conclusions

We have created a prototype web application that allows OBO Foundry ontology developers to see which classes from their ontologies overlap with classes from other ontologies in the OBO Foundry (http://obomap.bioontology.org webcite). From our analysis, we conclude that while the OBO Foundry has made significant progress toward orthogonality during the period of this study through increased adoption of explicit term reuse, a large amount of overlap remains among these ontologies. Furthermore, the characteristics of the identified overlap, such as the terms it comprises and its distribution among the ontologies, indicate that the achieving orthogonality will be exceptionally difficult, if not impossible.