Fig. 2
NetMHCpan Binder Predictions. a Using our experimental peptide sequences as inputs into NetMHCpan4.0 to predict epitopes for 22 HLA types represented in the 6 HIV-1 Infected people, a heatmap showing absolute counts of computationally predicted 9-mer binders against HIV-1 genes was constructed. The dendrogram shows the nearest similarity for the number of predicted counts across HLA types; b the length of the HIV-1 protein sequence plotted against the absolute number of NetMHCpan4.0 predicted 9mer binders showing a positive correlation (Spearman’s correlation coefficient, rs = 0.88). The number of distinct predictions is dependent on the length of the HIV-1 sequence; c comparison of HIV-1 clade A and D absolute number of NetMHCpan4.0 predicted 9mer binders per HIV-1 gene for the wet experiment test peptide sequences. The algorithm predicted more binders for clade D than clade A