This engine provides the information content (IC) of GO terms, semantic similarity between GO terms and GO-based protein functional similarity scores. The specificity of GO terms and the similarity of biological content between GO terms or proteins are transformed into numeric values for protein analyses at the functional level. This enables protein comparisons based on their biological roles and facilitates efficient exploitation of biological knowledge embedded in the GO DAG. These semantic similarity approaches have now been deployed in the GO enrichment (GOSS-FEAT), gene search (GOSP-FIT), functional classification (GOSP-FCT) and annotation prediction tools.