Dr. Robert Stevens is a senior lecturer in Bioinformatics in the bioHealth Informatics Group (http://www.cs.manchester.ac.uk/bhig) with a background in both biochemistry and computer science. His D.Phil in human computer interaction led to work focusing on biologist users of computer technology. He led the Life Science work on the TAMBIS and GONG (http://gong.manchester.ac.uk) projects, and has been instrumental in bridging between biologists and computer scientists in the myGrid project. He has an international reputation in ontology development and controlled annotation, including: early development on semantic similarity measures in bioinformatics; collaborations with the NERC environmental biology community; the BioPAX consortium; and members of the Gene Ontology Consortium; ontologies to map evolutionary relationships between species in ComparaGrid ( http://www.comparagrid.org/ ); and the development of annotation interfaces for working biologists (GOAT - http://goat.man.ac.uk & myTea - http://mytea.org.uk/) and a Life Science Semantic-based browser in the EU funded Sealife project ( http://www.biotec.tu-dresden.de/sealife/). Robert is a CI on the CO-ODE project ( http://www.co-ode.org/) that co-develops the Protege OWL ontology development environment. He co-chairs the databases and ontologies track at ISMB 2007 and has co-chaired the annual bio-ontologies meeting at ISMB for eight years. He has presented many tutorials on ontology and workflow development.
In this talk, I will describe recent work on using ontologies to help classify members of the protein phosphatases in a genome. Classification of proteins expressed by an organism is an important step in understanding the molecular biology of that organism. Traditionally, this classification has been done by human experts and it is regarded as the gold standard method. Human knowledge can recognise the properties that are sufficient to place an individual gene product into a particular protein family group. Automation of this task usually fails to meet this gold standard because of the difficult recognition stage. The need to automate the classification process by making human knowledge accessible in computational form is motivated by the growing number of genomes, the rapid changes in knowledge and the central role of classification in the annotation process. We capture human understanding of how to recognise members of the protein phosphatase family by domain architecture as an ontology. By describing protein instances in terms of the domains they contain, it is possible to use description logic reasoners and our ontology to assign those proteins to a protein family class. We have tested our system on classifying the protein phosphatases of the human and Aspergillus fumigatus genomes and found that our knowledge-based, automatic classification matches that of the human curators and for these two species we have also found putative new phosphatase proteins. We have made the classification process fast and reproducible and, where appropriate knowledge is available, the method can potentially be generalised for use with any protein family.
Much has been written of the facilities for ontology building and reasoning offered for ontologies expressed in the Web Ontology Language (OWL). Less has been written about how the modelling requirements of different areas of interest are met by OWL-DL's underlying model of the world. Just as small an amount has been written about how to best exploit what is possible to say in OWL. In this tutorial I will use the disciplines of biology and bioinformatics to reveal the requirements of a community that both needs and uses ontologies. I will use a case study of building an ontology of protein phosphatases to show how OWL-DL's model can capture a large proportion of the community's needs. I will demonstrate how ontology design patterns can extend inherent limitations of this model. I will give examples of non-binary relationships, lists and exceptions, and I will conclude by illustrating what OWL-DL, the proposed OWL 1.1 extensions and its underlying description logic either cannot handle in theory or because of lack of implementation. Finally, I will present an ontology building methodology called normalisation that not only follows some perceived best practice, but also allows OWL's rich potential to be exploited with relative ease.
Our work on using an OWL ontology to classify members of the phosphatase protein family has given us the prospect of cataloging innumerable proteins from many families and suggesting many new types of proteins. The barrier to really exploiting the potential is the development of the hundreds of protein family ontologies.
In this talk I'll raise the question of the ontology counterpart of the myExperiment idea; that is, "myOntology". Can we engage the community in building such ontologies? The lure might be the discovery of novel proteins and the means to analyse them through the community effort and social organisation of myExperiment. Can we gather information from tags supplied during analyses on proteins to help develop protein family ontologies? Protein family analysis through OWL, the bioinformatics analysis of the results and the commmunity development of the ontologies offer an ideal marketplace for eScience and ontologies.