MULTIPLE STRUCTURE ALIGNMENT BASED ON GEOMETRICAL CORRELATION OF SECONDARY STRUCTURE ELEMENTS

Protein structure alignment facilitates the analysis of protein functionality. Through superimposed structures and the comparison of variant components, common or specific features of proteins can be identified. Several known protein families exhibit analogous tertiary structures but divergent primary sequences. These proteins in the same structural class are unable to be aligned by sequence-based methods. The main objective of the present study was to develop an efficient and effective algorithm for multiple structure alignment based on geometrical correlation of secondary structures, which are conserved in evolutionary heritage. The method utilizes mutual correlation analysis of secondary structure elements (SSEs) and selects representative segments as the key anchors for structural alignment. The system exploits a fast vector transformation technique to represent SSEs in vector format, and the mutual geometrical relationship among vectors is projected onto an angle-distance map. Through a scoring function and filtering mechanisms, the best candidates of vectors are selected, and an effective constrained multiple structural alignment module is performed. The correctness of the algorithm was verified by the multiple structure alignment of proteins in the SCOP database. Several protein sets with low sequence identities were aligned, and the results were compared with those obtained by three well-known structural alignment approaches. The results show that the proposed method is able to perform multiple structural alignments effectively and to obtain satisfactory results, especially for proteins possessing low sequence identity.