An Information Theoretic Approach for the Analysis of RNA and DNA Binding Sites

Proteins perform several critical biological processes by interacting with other macromolecules (DNA, RNA) and small molecules. Several computational approaches have been developed to determine the protein interaction sites using sequence and structure features. Instead of building another adhoc prediction algorithm, the purpose of this study is to understand the contribution of a protein’s residue in a RNA-binding event and compare it with the DNA-binding process. We evaluate several sequence and structure-based features using mutual information theory. We show that solvent accessibility and profile-based features can be used for developing good protein-RNA binding site determination algorithms. We also recommend features that could discriminate between RNA and DNA binding sites. This work can be extended to understand protein-protein and protein-ligand interactions as well.