Phenotype ontologies and translational research (Cambridge / Aberystwyth)

Overview

From 2010 to 2014 Robert held postdoctoral positions on the Wellcome Genome Campus near Cambridge — first at the EBI and then at the Wellcome Trust Sanger Institute in the team led by Christopher Mungall and Damian Smedley's mouse-phenotype work — followed by a lectureship at Aberystwyth University. The unifying theme across both posts was building the infrastructure to use phenotype ontologies at scale for both basic and translational biology, and demonstrating that this infrastructure changed what was possible in gene-function prediction and rare-disease diagnosis.

The principal deliverable of the Cambridge years was PhenomeNet, a cross-species phenotype network in which gene and disease entities from human, mouse, zebrafish, fly, worm, yeast, and other model organisms were linked by automatically computed semantic similarity between their phenotype profiles. PhenomeNet provided a single substrate on which a mouse knockout phenotype, a human Mendelian disorder, and a yeast deletion strain could be compared on equal footing, and it became the engine behind phenotype-driven disease-gene prioritization tools used by the Sanger mouse-genetics programme, the Monarch Initiative, and the Exomiser variant-prioritization pipeline used in the 100,000 Genomes Project. The Aberystwyth lectureship period extended this with AberOWL, an ontology repository and reasoning service that exposed hundreds of biomedical ontologies through a description-logic query interface — unlike BioPortal or OLS at the time, AberOWL ran an OWL EL reasoner over its content, so users could ask for subsumees of arbitrary class expressions and get answers grounded in the ontologies' axioms rather than only in their asserted hierarchies.

Around these two pieces of infrastructure, the group at Aberystwyth produced a set of translational applications: phenotype-based prioritization for whole-exome and whole-genome sequencing in rare disease, semantic similarity for drug repurposing, and ontology-based representations of mouse phenotyping data to support large-scale international projects (IMPC, MGI, ZFIN). The Aberystwyth period also delivered the first generation of group members — PhD students and research associates — several of whom continued into the KAUST group after 2014. By the end of this period, phenotype-driven, ontology-reasoning-based methods had moved from a methodological curiosity to a routine component of clinical and model-organism genomics pipelines, and the tools developed in Cambridge and Aberystwyth remain part of that pipeline today.

Period: 2010–2014

Software