Overview of the methodology. (a) Annotations from several sources are assigned to genes in the input list. (b) The apriori algorithm is applied to find sets of annotations that frequently co-occur in the input list. (c) The statistical significance of each annotation or set of concurrent annotations is calculated based on its frequency in the input and reference sets. The figure illustrates an example in which a list of yeast genes is annotated with Gene Ontology (GO) terms for 'cellular component' and KEGG pathways. In the output table only the annotations that co-occur in more than five genes are shown.