A new algorithm for sequential and non-sequential protein multiple structure alignment

A fundamental issue in studies of protein structure and function is how to identify the conserved common core in multiple protein structures. The existing algorithms work well for proteins family such as those in HOMSTRAD but are not satisfactory for general multiple structure alignment problems, especially in some challenging cases, such as the occurrence of circular permutations. In this paper, an efficient approach called SANA-mult for multiple structure alignment is presented. Specifically, the alignment problem is first mapped on to a mixed integer programming problem via introducing a structural template, and then the algorithm decomposes the problem into two subproblems, i.e. solving the pairwise alignment and updating the template chain. We show that the proposed method can obtain sequential and non-sequential solutions for multiple structure alignment in an accurate manner, which is competitive or superior to the existing methods. The effectiveness of the new algorithm SANA-mult is tested using various protein structure sets and benchmark examples.