Prediction of proteasomal cleavage sites using PCPS

The proteasome complex is the main responsible of proteolytic degradation of cytosolic proteins, generating the C-terminus of MHC I-restricted peptide ligands and CD8 T cell epitopes. Prediction of proteasomal cleavage sites is thus relevant for predicting CD8 T cell epitopes. We previously reported an $n$ -gram based method to predict proteasomal cleavage sites named PCPS, which is implemented for free public use online at http://imed.med.ucm.es/pcps/. Here, we used a set of 59 Hepatitis C Virus (HCV) CD8 T-cell epitopes to analyze cleavage predictions by PCPS and compared them with those provided by a related method implemented in NetChop web server. PCPS clearly outperformed NetChop in sensitivity (0.88 and 0.78, respectively); it identified much better than NetChop that the C-terminus of tested CD8 $T$ cell epitopes likely resulted from proteasomal cleavage. However, PCPS identified additional preferential cleavage sites within the tested epitopes more often than NetChop, which resulted in lower specificity (0.57 vs 0.66, respectively). Proteasomal cleavage site predictions are used to enhance CD8 T cell epitopes by discarding peptides resulting from peptide-MHC I binding models that do not have a C-terminus compatible with proteasome production. Thereby, we have now implemented proteasomal cleavage site predictions provided by PCPS $n$ -grams in RANKPEP, a tool to identify peptides presented by MHC molecules, allowing to filter out peptides predicted to bind to MHC I molecules that are unlikely to result from proteasomal cleavage. RANKPEP is available for free public use online at http://imed.med.ucm.es/Tools/rankpep_new.html.