A functional ontology-based score for template-based protein docking

Comparative docking generates structures of protein-protein complexes using known structures as the templates for modeling. The detection of templates is often based on structural similarity between the modeling target complex and the experimentally determined complexes. The performance of the template-based docking significantly decreases when the templates are only moderately similar to the target. We developed a non-structural score, which is complementary to the structural similarity scoring, based on the target/template similarity of the Gene Ontology annotations. A combined scoring function improves discrimination of wrong templates, enhancing the reliability of the docking.