PASS2: update of database of structure-based sequence alignments

Protein sequence alignments are evolutionary models and offer as starting points for the recognition of additional members of a homologous family and design of experiments. However, the accuracy of sequence alignments is obscured at the superfamily level due to distant relationships. Where structures of proteins are available, distantly related proteins can be aligned, guided by structural features. The Protein Alignment Organized as Structural Superfamilies (PASS2) database offers such structure-based sequence alignments for protein domains classified within superfamilies, as per the Structural Classification of Proteins extended (SCOPe) framework. The present update of PASS2 (PASS2.8) corresponds to the latest SCOPe release (version 2.08). This release comprises data for 26 690 protein domains exhibiting less than 40% sequence identity, organized into 2058 superfamilies. Several features derived from these alignments, including conserved secondary structural motifs, hidden Markov models (HMMs), conserved residues, and interactions across superfamilies, are also provided. For superfamilies containing divergent members, a k-means clustering algorithm has been employed to identify outliers and partition domains into split superfamilies. Novel features in this update include topological diagrams of the domains, potential interactors for each domain, and an updated methodology for identifying conserved interactions across superfamilies. This version of the database can be reached from http://caps.ncbs.res.in/pass2.