Abstract:

The human leukocyte antigen (HLA) system is an important part of the immune system involved in the presentation of antigen fragments (oligopeptides) to the T-cells.The proteins encoded by the HLA class II genes of locus DP are associated with a significant number of autoimmune diseases, as well as with the susceptibility or resistance to a number of infectious agents. The aim of the present study is to analyse the structure – affinity relationships of antigen peptides binding to 7 most common in the human population HLA-DP proteins. The analysis is performed by proteochemometrics. A set of 3,864 15-mer peptides, known binders and non-binders to HLA-DP proteins, are compiled from IEDB. The set is pre-processed and divided into training (80%) and test (20%) subsets. The training set is used to derive a proteochemometric model by iterative self-consistent partial least squares-based algorithm. The derived model has good explaining capacity (R² = 0.899 and Q² = 0.892), but moderate predictive ability validated by the test set (R_pred = 0.515). The proteochemometrics is a suitable method for structure-affinity analysis of peptides binding to multiple HLA proteins.