Three Dimensional Protein Structure Alignment Algorithms with Selective Secondary Structure Information

Abstract. The knowledge of structural classes is applied in numerous important predictive tasks that address struc-tural and functional features of proteins. In our previous work [17], algorithms are proposed to refine the given align-ment by stepwise finding better alignments of the protein pairings using minimum bipartite matching method on ge-ometric distance space and several other adjustment strate-gies. The proposed methods are used to improve the given initialized alignment of two structures. The refinement method is applied to improve several previous known meth-ods including VAST [7] and CE [18]. In this paper, we propose methods of combining differ-ent segmentation methods that incorporate the secondary structural information. The experimental results show that our method improves the aligned rmsd value obtained from the VAST package by an averaged improvement ratios about 18%. The best quality is obtained by combining different sources of secondary structural information. These set-tings are later adjusted by local refinement algorithm with trigonometric series estimation. Here we also propose ways of balancing two constraints of the time resource as well as the quality requirement. Systematic ways of selecting appro-priate algorithms according to different users ’ quality/time preference are addressed. We show that it is possible to adjust the parameters to get the result that are sufficiently satisfactory and yet spent much less time resource. Further-more, some of our preliminary result is accessible through the web interface