Molecular Docking Analysis of 120 Potential HPV Therapeutic Epitopes Using a New Analytical Method

Abstract

The accurate modelling and scoring of protein–peptide (Pr–Pe) complexes are determining factors in the drug discovery studies by molecular docking. However, there are three main problems with Pr–Pe docking methods: high flexibility of protein and peptide molecules, scoring of generated models, and integrative modelling. In this study, we introduced a new analytical method for ranking epitopes to reduce scoring and integrative modelling problems. At first, we docked 120 potential epitopes from E5, E6 and E7 oncoproteins from human papillomavirus (HPV) 16, 18, 31 and 45 against ten HLA-I and six HLA-II molecules using two servers, CABS-dock (a template-free method (TFM)) and GalaxyPepDock (a template-based method (TBM)). Then, we weighted the early (original) docking scores by involving the population coverage (PC) of the HLA molecules and integrated all the weighted scores earned by each peptide to calculate a single weighted-integrated (W-I) docking score per each peptide to rank them. Our findings showed that there are positive bivariate correlations between the results of applied servers, with Pearson correlation coefficient averages (r) 0.385 and 0.525, as well as, R square averages (r2) 0.148 and 0.275 for the original and W-I scores, respectively (p < 0.01). These statistics indicated that the W-I scores, analyzed by the new method, have more correlation and predictability power than original scores. Generally, this new analytical method can be considered as an effective solution to improve the ranking of epitopes in similar studies to avoid obtaining more false results or even true results but without appropriate PCs. Additionally, our high-score epitopes are valuable candidates for further experimental validating studies, targeting the treatment of HPV-mediated cancers.