Last month I participated in an informal meeting for users and developers of the Phenotype and Trait Ontology (PATO), organized by Chris Mungall and Suzanna Lewis at the Lawrence Berkeley National Labs in Berkeley, California. PATO is the quality ontology used by many in the OBO community to annotate phenotypic variation. We are heavy-duty PATO users here at Phenoscape, so I was eager to meet with its developers and other users to discuss outstanding issues and hear about phenotype annotation from other projects.
The two-day meeting brought together diverse PATO users with the goals of establishing best practices for phenotype annotation, familiarizing everyone with the Phenote annotation tool, and discussing strategies for decomposing terms from the Human Phenotype Ontology into Entity-Quality syntax. Participants at the meeting were Sandra Dölken (Charité, Universitätsmedizin Berlin, Human Phenotype Ontology with Peter Robinson), George Gkoutos (University of Cambridge, PATO gatekeeper), Melissa Haendel (ZFIN, data curator and ZFA maintainer), Nomi Harris (BBOP), Sarah Maynard (UCSD, Neuroscience Information Network – NIF, and Protein Knowledgebase Ontology – PKB), Martin Reese (Omicia), Paul Schofield (University of Cambridge, PATHbase and CASIMIR), and Nicole Washington (BBOP).
After project introductions, we started off with software demos and heard about an interesting mix of stand-alone and web-only tools. Sandra demonstrated use of the online Phenomizer and Disease Annotator tools used by clinicians, I demoed the Phenoscape annotation tool (Phenex), and Chris demonstrated queries on Sarah’s annotations loaded into OBD. Sarah also introduced us to Neurolex.org, a wiki-based resource containing NIF ontologies that can be edited by users, with contributions reviewed and added to the ontologies by curators. We then had a discussion of general annotation guidelines, in which I ran through the annotation guidelines we’ve developed at Phenoscape. This led to discussion about unresolved annotation issues including the various ways of representing size, how to refer to one member of a pair (e.g., left and right paired bones), and how to represent an individual of a series.
Following the discussion of curation guidelines, we began the paired-annotation drill of disease phenotypes from OMIM (Online Mendelian Inheritance in Man). We used a special configuration of Phenote that contains a “disease” field for the relevant OMIM disease; “gene” field for known genes; “phenotype” field for terms from HPO, entity field for terms from the Foundational Model of Anatomy, quality field for PATO and Spatial Ontology terms, and free text fields for Age of Onset and Frequency. In addition to establishing curation guidelines for HPO, this exercise resulted in many requests for new FMA terms, PATO terms and synonyms, and feature requests for Phenote.
The relatively short meeting was a good opportunity to discuss the different approaches and commonalities to representing complex phenotypes. The meeting wrapped up with a great meal at Trattoria Corso, followed by a long list of action items prepared by Suzie after everyone returned home.