AutoPSI: a database for automatic structural classification of protein sequences and structures

In protein research, structural classifications of protein domains provided by databases such as SCOP play an important role. However, as such databases have to be curated and prepared carefully, they update only up to a few times per year, and in between newly entered PDB structures cannot be used in cases where a structural classification is required. The Automated Protein Structure Identification (AutoPSI) database delivers predicted SCOP classifications for several thousand yet unclassified PDB entries as well as millions of UniProt sequences in an automated fashion. In order to obtain predictions, we make use of two recently published methods, namely AutoSCOP (sequence-based) and Vorolign (structure-based) and the consensus of both. With our predictions, we bridge the gap between SCOP versions for proteins with known structures in the PDB and additionally make structure predictions for a very large number of UniProt proteins. AutoPSI is freely accessible at http://www.bio.ifi.lmu.de/AutoPSIDB.