Biological Feature Incorporated Alignment for Cross Species Analysis on Carbohydrate Binding Modules

Multiple sequence alignment is widely applied to discover core conserved regions among query sequences. However, the major deficiency is that alignment accuracy is extremely sensitive to primary sequence identity, which causes alignment of low identity sequences difficult. We propose a feature-integrated model called feature-incorporated alignment (FIA) which integrates relevant biological characteristics including aromatic amino acids, hydrophilicity, beta-stranded structure, and BLOSUM62 matrix to locate ligand-binding residue in carbohydrate binding modules (CBMs), a protein family with fairly low sequence identify but highly functional correlation. The results indicated that FIA can not only detect aromatic residues on the outer surface of structure, but also achieve better accuracy than ClustalW2 and DIALIGN-TX on entropy criterion in all three test datasets from CBMs. Computational analysis in CBMs can facilitate the discovery of crucial ligand-binding residues of carbohydrate-active enzymes.