Protein Structure Alignment Based On Secondary Structure Elements

Protein structure alignment, which is valuable for understanding the evolutionary relationships between the proteins and discovering the mechanism of protein folding, is one of the most important research issues in bioinformatics. Many alignment algorithms based on secondary structure elements (SSEs) have been developed, which are advanced in cost-effectiveness, but the performance of these methods still can be improved.In this study, a simple and effective superposition algorithm is used as an elementary component. By combining several geometrical features and the hydrophobicity characteristic, a structural similarity scoring strategy is designed and used in the dynamic programming process of aligning the SSEs. Aiming at the problem that the scoring method can not well combine SSEs similarity with the number of aligned residues, a correction function is joined with the similarity score. In residue alignment process, a method based on dynamic programming is used to extract a subset which is aligned in ascending order from the nearest-neighbors pairs of all residues, reducing the probability of gaining a local optimized result. On the principle of expanding the search space, a change is made in the way of combining SSEs and residues alignment processes. All possible original transformations between structures are searched thoroughly. Meanwhile, measures are taken to guarantee the efficiency of this algorithm. A new program SECS is developed based on the improvements above. Testing results on 1054 proteins from database FSSP and other similarity datasets demonstrate that the performances of SECS have an average 9.33% advantage in RMSD, and an average 2.60% advantage in the number of aligned residues compared to the well-known algorithm LOCK2.