Abstract
Analysis of functiona/structure relevant residues in proteins can lead to the prediction of new functions of proteins and identify the core structure of proteins. To this end, we calculate the relative entropy from the multiple sequence alignments of the sequences with the same SCOP fold classification, PDZ domain-like, SH3 domain-like, and protein kinase-like. With this method, we identify some relevant residues that have been known important in function or in structure and also find some residues possibly important in stabilizing structure that help the key residues bind to the ligand. Our work shows that relative entropy is a useful measure for analyzing the relevant residues in proteins even though the positions are not highly conserved.