Learning from ontological annotation : an application of formal concept analysis to feature construction in the gene ontology

A key role for ontologies in bioinformatics is their use as a standardised, structured terminology, particularly to annotate the genes in a genome with functional and other properties. Since the output of many genome-scale experiments results in gene sets it is natural to ask if they share common function. A standard approach is to apply a statistical test for overrepresentation of ontological annotation, often within the Gene Ontology. In this paper we propose an alternative to the standard approach that avoids problems in over-representation analysis due to statistical dependencies between ontology categories. We use a feature construction approach to pre-process Gene Ontology annotation of gene sets and incorporate these features as input to a standard supervised machine learning algorithm. Our approach is shown to allow the straightforward use of an ontology in the context of data sourced from multiple experiments to learn a classifier predicting gene function as part of cellular response to an environmental stress.