Journal of Proteomics & Bioinformatics
Open Access

Abstract

Meta-analysis of Protein Structural Alignment

Jim Havrilla and Ahmet Saçan

The three-dimensional structure of a protein molecule provides significant insight into its biological function. Structural alignment of proteins is an important and widely performed task in the analysis of protein structures, whereby functionally and evolutionarily important segments are identified. However, structural alignment is a computationally difficult problem and a large number of heuristics introduced to solve it do not agree on their results. Consequently, there is no widely accepted solution to the structure alignment problem. In this study, we present a meta-analysis approach to generate a re- optimized, best-of-all result using the alignments generated from several popular methods. Evaluations of the methods on a large set of benchmark pairwise alignments indicate that TMalign (Template Modeling Alignment) provides superior alignments (except for RMSD, root mean square deviation), compared to other methods we have surveyed. Smolign (Spatial Motifs Based Multiple Protein Structure Alignment) provides smaller cores than other methods with best RMSD values. The re-optimization of the alignments using TM-align’s optimization method does not alter the relative performance of the methods. Additionally, visualization approaches to delineate the relationships of the alignment methods have been performed and their results provided