Druggability Analysis and Prediction Based on Geometric Distances between Amino Acid Residues and Protein Surface Pockets

Protein is the major component of the organism. A concave (pocket) on a protein surface is known to be the best target for a drug to react. We previously presented a study on distance analysis between pockets and amino acid residue. We firstly identified pockets on the protein surface and then calculated distances between atoms of an amino acid residue and the deepest points or the outer loops of the pockets. We extracted proteins which at least one of the pockets are close to arbitrary pairs of amino acid residues, calculated the ratios of druggable proteins, and visualized the distribution of the ratios as a colored matrix. We suggested from the visualization results that particular pairs of amino acid residues may affect the druggability of the proteins in our previous study. This paper presents an extension of our study to explore the relevance between druggability of proteins and distances between a set of amino acid residues and protein surface pockets. Our technique treats the pockets as 20-dimensional vectors consisting of distances to each of amino acid residues, and applies GeodesicSOM with the set of the vectors. Spherical maps generated by GeodesicSOM are used to visualization of distribution of the pockets in the 20-dimensional vector space, and estimation of druggability of proteins with the 20-dimensional vectors of the pockets.