Abstract
With the increasing number of protein structures in Protein Data Bank(PDB) efficient and accurate algorithms similar to local alignment of sequences (BLAST) are necessary to classify newly discovered structures into appropriate super families and also to identify structural motifs. Several structural alignment algorithms have been proposed in the literature (DALI, CE, PSIST, TM-Align, etc.). Most of these algorithms perform a global structural alignment between pairs of structures and provide normalized scores which can help in classification of the structures. However they donpsilat address the problem of identification of structural motifs. In this paper we present efficient algorithms for local structural alignment based on a new idea of using Variable Length Alignment Fragment Pairs (VLAFPs) in contrast to Constant Length Alignment Fragment Pairs (CLAFPs) used by all the existing algorithms such as CE, TM-Align, DALI and PSIST. Our VLAFP algorithm is independent of the scoring schemes used to score the CLAFPs and can work with any scoring scheme like the TM-Score. We also introduce two new scoring schemes based on center of gravity of the atoms. Experiment results indicate that using VLAFPs can improve both the quality of local structural alignment and also accuracy of the classification. Using our VLAFP algorithm together with our new scoring schemes we could acheive an average super family accuracy of 84% and a class accuracy of 87% on structures with very less sequence homology.
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have