Functional Residue Prediction by Multiple Sequence Alignment for Carbohydrate Binding Modules

Abstract

Multiple sequence alignment is often used to locate consensus sequence stretches with evolutionary and functional conservation. However, when sequence similarity among the queries becomes low, sequence alignment tools generate extremely diverse results. The aim of this study is to incorporate relevant biological knowledge and assumptions to improve quality of general alignment on low similarity sequences. Since recognition of key features in carbohydrate binding module (CBM) family does not apply to general models, a more accurate weighted entropy function employing secondary-structure-based and key-residueweighted algorithms for alignment was designed to approach this goal. The results indicate that the proposed method is able to detect the known ligand-binding residues and to predict unknown functional residues in cellulose binding domains (CBDs) and xylooligosaccharides binding domains (XBDs) in terms of three-dimensional structures. Our results contribute molecular basis of CBDs and XBDs and potential application in development of alternative energy for future needs.