A set of new amino acid descriptors applied in prediction of MHC class I binding peptides

Abstract A set of new amino acid descriptors, namely factor analysis scales of generalized amino acid information (FASGAI) involving hydrophobicity, alpha and turn propensities, bulky properties, compositional characteristics, local flexibility and electronic properties, was proposed to resolve the representation of peptide structures. FASGAI vectors were then used to represent the structures of 152 HLA-A ∗ 0201 restrictive T-cell epitopes with 9 amino acid residues. The features that are closely related to binding affinities were selected by genetic arithmetic, and the model based on partial least squares was developed to predict binding affinities. The model revealed promising predictive power, giving relatively high predictions for training and test samples. Further, the PreMHCbinding program at significantly lower computational complexity was exploited to predict MHC class I binding peptides. Quantitative structure–affinity relationship analyses demonstrated the bulky properties and hydrophobicity of the 3rd residue, bulky properties of the 2nd residue, hydrophobicity of the 9th residue that provided high positive contribution to the binding affinities, and that the hydrophobicity of the 4th residue and local flexibility of the 3rd residue were negative to binding affinities. The results showed that FASGAI vectors can be further utilized to represent the structures of other functional peptides; moreover, it has thus showed us further direction into the potential applications on relationship between structures and functions of proteins.