A structure-based, quantitative structure-activity relationship approach for predicting HLA-A*0201-restricted cytotoxic T lymphocyte epitopes.

In this investigation, we first constructed four types of non-bonding interaction matrixes by defining direct contacting residue types for HLA-A*0201 protein in interaction with each position of HLA-A*0201-restricted cytotoxic T lymphocyte epitope as well as several formulae calculating ligand/receptor non-bonding interactions. Relative to these studies, a method which we refer to as structure-based, quantitative structure-activity relationship is proposed and utilized for studies on 266 HLA-A*0201-restricted cytotoxic T lymphocyte epitopes. The resulting genetic algorithm-partial least square regression model is consistent with both published studies and molecular graphics analysis. Two non-bonding interactions (i.e. hydrophobic and hydrogen bonding), are found to play important roles in antigen recognition and presentation, especially exerting effects at positions of anchor residues in antigen peptides.