Submission 14

Submission 14

The Pistoia Alliance Ontologies Mapping (OM) project (http://www.pistoiaalliance.org/projects/ontologies-mapping) was created to find or make better tools or services for mapping between ontologies in the same domain and to establish best practices for ontology management in the Life Sciences. The project gives access to standardized tools, methodologies and service for mapping and visualisation of ontologies which supports more effective integration and analysis of data. One important output from the project is a publicly accessible set of ontology guidelines which can be used to as a “health check” prior to application (https://pistoiaalliance.atlassian.net/wiki/spaces/PUB/pages/43089942/Ontologies+Guidelines+for+Best+Practice). 

The project has focused on mapping between disease and phenotype ontologies. Currently, such mappings are generated mostly by curators which limits the efficiency and scalability of mapping, which can be improved by development of automated systems. The project has specified the requirements for ontology mapping in terms of graphical user interface, matching algorithm, workflow environment and performance. This has enabled the systematic evaluation of existing OM tool systems to identify those with top capabilities and mapping performance from commercial and academic providers. The ontology matching algorithm is critical to mapping performance, so we have organised the phenotype mapping tasks for the annual Ontologies Alignment Evaluation Initiative (http://oaei.ontologymatching.org). Top performing OM algorithms have been identified through this process which originate mostly from academic groups. 

A prototype OM service has been developed to provide and evaluate the quality mappings between public ontologies. Such a service is required because source ontologies change regularly so any mapping has to be maintained to remain current. We are making full use of the existing ontology services at EMBL-EBI who have joined our project. This has enabled the efficient development of an algorithm for mapping predictions which are being evaluated and stored in a central repository for ontology mappings, named OxO (https://www.ebi.ac.uk/spot/oxo).