Abstract
3D-QSAR of peptides is a daunting task. The difficulty in peptide QSAR arises due to the sheer number of conformational degrees of freedom for peptides that makes alignment in a 3D grid an overwhelming task. In this paper, we propose a method of QSAR where the alignment of peptides is shifted from 3D space to 1D space, making the alignment of peptides a very simple proposition. The method called HomoSAR, is based on an integrated approach that uses the principles of homology modeling in conjunction with the QSAR formalism to predict and design new peptide sequences. The peptides to be studied are subjected to a multiple sequence alignment which is followed by scoring every position in the peptide sequence against a reference peptide in the alignment, through calculation of similarity indices. The
similarity indices obtained for each position (amino acid residue) in the peptide form the “descriptor” values (independent variables) which are then correlated to the biological activity of the peptide by G/PLS techniques. As an application, the methodology has been illustrated for the dataset of nonamer peptides that bind to the Class I major histocompatibility complex (MHC) molecule HLA-