Tools and Algorithms for Protein Structures Comparison with Various Initial Con¯gurations

Abstract

Comparison of protein structures pro- vide the opportunity to recognize homology that is undetectable by sequence comparison, and it rep- resents a powerful means of discovering functions, yielding direct insight into the molecular mech- anisms. Currently, there are several techniques available in attempting to ¯nd the optimal align- ment of shared structural motifs between two pro- teins. In this paper, we propose algorithms and develop tools for local alignment between two protein struc- tures by means of local adjustment. In our pre- vious work [18], we show that the trigonometric series approximation is appropriate for estimating the good isometric transformations of one struc- ture and aligning it to the other structure. Based on these results, here we propose algorithms to re- ¯ne the given alignment by stepwise ¯nding better alignments of the protein pairings using minimum bipartite matching method on geometric distance space and several other adjustment strategies. The proposed methods are used to improve the given initialized alignment of two structures. Furthermore, we also propose several prelimi- nary initialization algorithms to examine the ef- fectiveness of the proposed local re¯nement algo- rithms. We show the e®ectiveness of the proposed re¯nement methods and initial algorithm by a set of experiments, which improve several previous re- sults. Furthermore, some of our preliminary result is accessible through the web interface.