Iterative Refinement Techniques for Improving Constrained Feature Based Multiple Structure Superposition

Abstract

As large amount of protein tertiary structures are available, an efficient and effective tool for protein structural alignment becomes important for the detection and classification of protein function and fold recognition. In this paper, we present a novel iterative refinement algorithm for improving the multiple protein structure superposition based on constrained features of protein sequences. This algorithm is achieved according to its initial alignment from conserved motifs, and an updated virtual center structure is created iteratively followed by optimal pairwise alignments between each query protein and the updated virtual center structure. The refined alignment converges to its best alignment results based on the measurement and identification of the minimal average root mean square deviation (RMSD) and maximum number of average aligned residues. The effectiveness of the proposed algorithm was verified by comparing the performance between the constrained features based multiple structure alignment with or without adding the extra iterative refinement techniques.