Peptide-MHC class I binding affinity predictions.
FASTA-formatted input protein sequences from the entire SARS-CoV-2 and SARS-CoV proteomes were obtained from the NCBI RefSeq database (67) under accession numbers NC_045512.2 and NC_004718.3. We kmerized each of these sequences into 8- to 12-mers to assess MHC class I-peptide binding affinity across the entire proteome. MHC class I binding affinity predictions were performed using 145 different HLA alleles for which global allele frequency data were available as described previously (72) (see Table S5) with netMHCpan v4.0 (73) using the ‘-BA’ option to include binding affinity predictions and the ‘-l’ option to specify peptides 8 to 12 amino acids in length (Table S1). Binding affinity was not predicted for peptides containing the character ‘|’ in their sequences. Additional MHC class I binding affinity predictions were performed on all 66 MHCflurry-supported alleles (–list-supported-alleles; Table S6) using both MHCnuggets 2.3.2 (74) and MHCflurry 1.4.3 (75) (see Tables S7, S8, and S9 and Fig. S7 to S10 in the supplemental material). We further cross-referenced these lists of peptides with existing experimentally validated SARS-CoV epitopes present in the Immune Epitope Database (Table S4) (76). We then performed consensus binding affinity predictions for the 66 supported alleles shared by all three tools by taking the union set of alleles and filtering for peptide-allele pairs matching the union set of alleles. For the SARS-CoV-specific and SARS-CoV-2-specific distributions of per-allele proteome presentation, we exclude all peptide-allele pairs with >500 nM predicted binding. In all cases, we used the netchop v3.0 (77) “C-term” model with a cleavage threshold of 0.1 to further remove any peptides that were not predicted to undergo canonical MHC class I antigen processing via proteasomal cleavage (of the peptide’s C terminus).