Multiple structural alignment for distantly related all b structures using TOPS pattern discovery and simulated annealing

Abstract

Topsalign is a method that will structurally align diverse protein structures, for example, structural alignment of protein superfolds. All proteins within a superfold share the same fold but often have very low sequence identity and different biological and biochemical functions. There is often signi®cant structural diversity around the common scaffold of secondary structure elements of the fold. Topsalign uses topological descriptions of proteins. A pattern discovery algorithm identi®es equivalent secondary structure elements between a set of proteins and these are used to produce an initial multiple structure alignment. Simulated annealing is used to optimize the alignment. The output of Topsalign is a multiple structure-based sequence alignment and a 3D superposition of the structures. This method has been tested on three superfolds: the b jelly roll, TIM (a/b) barrel and the OB fold. Topsalign outperforms established methods on very diverse structures. Despite the pattern discovery working only on b strand secondary structure elements, Topsalign is shown to align TIM (a/b) barrel superfamilies, which contain both a helices and b strands.