Over the years, a number of different vertebrate anatomy ontologies have been developed. Some of these are dedicated to a single model species, or to human. Others have been developed to describe phenotypic variation across species, and these cover a broad range of species. In particular:
- The Teleost Anatomy Ontology (TAO) [Dahdul et al 2010] covers teleosts (bony fish)
- The Amphibian Anatomy Ontology (AAO) [Maglia et al 2007] covers amphibians
- The Vertebrate Skeletal Anatomy Ontology (VSAO) [Dahdul et al 2012] covers the skeletal system of vertebrates
- The vertebrate Homologous Organs Group (vHOG) [Niknejad et al 2012] ontology describes structures with shared ancestry across vertebrates
- Uberon represents a range of anatomical structures across animals[Mungall et al 2012], with a content bias towards vertebrate anatomy
This lead to considerable duplication of effort, as common anatomical
structures such as ‘pectoral girdle‘ were represented in all five ontologies
(as well as their single species counterparts):
It was difficult for the Phenoscape group to integrate data across all these ontologies, as this required that curators kept mutual cross-references up to date, a time-consuming and error-prone task.
As a result, the maintainers of these ontologies agreed to join forces and build a common ontology.This work is described in a new paper in the ontologies special issue of the Journal of Biomedical Semantics:
Haendel MA, Balhoff JP, Bastian FB et al Unification of multi-species vertebrate anatomy ontologies for comparative biology in Uberon Journal of Biomedical Semantics 2014, 5:21 doi:10.1186/2041-1480-5-21
The group selected Uberon as the core ontology, as it had the broadest coverage, was already well-integrated with the single-species ontologies, and was adapted for OWL reasoning. The curators of these ontologies worked long and hard to integrate their work, with input from anatomy experts and developers of single-species ontologies, revealing many interesting differences in the way structures are represented across species along the way. For example, the representation of teeth in the combined ontology had to be flexible enough to accommodate teeth that are in widely variable locations and configurations:
The number of classes merged is shown in figure 2 of the paper:
As a result of this effort, we have a common anatomy ontology with broad and deep coverage for vertebrate anatomy. For a variety of viewing options, see the Uberon website. For examples of use for data integration see:
Like most ontologies, work is ongoing and we are constantly striving to improve depth, coverage and quality. We’re currently actively improving the representation of facial muscles in the ontology based on the FEED ontology. We are also working on a federated approach for bringing in invertebrate anatomy ontologies, many of which are developed under the auspices of the Phenotype RCN including the Arthropod Anatomy Ontology, the Poriferan anatomy ontology [Thacker et al, accepted, JBMS], the cephalopod ontology and the ctenophore ontology. We welcome feedback from everyone!