CAALIGN: a program for pairwise and multiple protein-structure alignment

Coordinate superposition of proteins provides a structural basis to protein similarity and therefore complements the technique of sequence alignment. Methods that carry out structure alignment are faced with the problem of the large number of trials necessary to determine the optimal alignment solution. This article presents a method of carrying out rapid (subsecond) protein-structure alignment between pairs of proteins based on a maximal C(alpha)-atom superposition. The algorithm can return alignments of 12 or more residues in length as multiple non-overlapping solutions of alignment between a pair of proteins which are independent of the fold connectivity and secondary-structure content. The algorithm is equally effective for all protein fold types and can align proteins containing no secondary-structure elements such as is the case when searching for common turn structures in proteins. It has high sensitivity and returns the set of true positive results before any false positives as judged by SCOP classification. It can find alignments between topologically different folds and returns information about sequence alignment based on structure alignment. Additionally, this algorithm has been extended to carry out multiple structure alignment to determine common structures within groups of proteins, including the nondegenerate set of proteins in the PDB. The algorithm has been implemented within the program CAALIGN and this article presents results from pairwise structure alignment, multiple structure alignment and the generation of common structure fragments found within the PDB using multiple structure alignment.