Similarities between proteins detected at the sequence and 3D-structure level reveal common or conserved functional properties as well as evolutionary relationships, and guide protein classifications. The Conserved Domain Database (CDD) and the Vector Alignment Search Tool (VAST) are resources supported by the National Center for Biotechnology Information (NCBI). CDD provides overlapping and partially redundant annotation on protein sequences; CDD models are built from representative protein sequence fragments, which are related by common descent, and similar models are clustered into protein domain superfamilies. We present a computational procedure to contrast classification via membership in conserved domain superfamilies with structural similarity as reported by VAST. We retrieve domain superfamilies that have 3D structure representatives and query those representatives for similarities of the corresponding 3D coordinate sets as reported by VAST. If such matches can be found, we consider the CDD superfamily clustering, which relies on significant sequence similarities, to be confirmed by the structure comparison method. This procedure can therefore be used to screen for �false clustering� in CDD superfamilies, highlighting cases where structure linked sequence fragments associated with a superfamily cluster fall into two or more distinct clusters of structural similarities according to VAST. We have examined around 5000 domain superfamilies from CDD and find that more than 85% of the domain superfamilies are directly supported by VAST. We determine a �threshold� for mutual structural similarity between domain superfamilies to unite existing domain superfamilies into even larger sets of domains with similar or conserved three-dimensional structure and these sets of domains are compared with the 'fold' annotation in SCOP (Structural Classification of Proteins) and/or CATH. The resulting sets are in good agreement (70-80%) with �fold� classifications in SCOP.

Testing the consistency of protein sequence classification with protein 3D-structure similarity

248th National Meeting (2014)

The official archive and repository for all ACS Meetings & Expositions abstracts and select posters or presentations.