Fig. 5: Variable predicted EBV antigen presentation underlies EBV DNA persistence.

a, Schematic of scoring the EBV proteome for antigen presentation against all HLA alleles observed in the UKB NFE cohort using NetMHCpan and NetMHCIIpan⁴⁸. b, Summary of top antigens bound per MHC class, coloured by whether the peptide is experimentally confirmed in IEDB⁵¹. c, Enrichment analyses of immunodominant peptides as a function of EBV gene functional class. The highlighted dot shows MHC I peptides are depleted for latency-associated genes. The significance cutoff was drawn at P = 0.05/4, to adjust for the four gene classes tested. Statistical test: two-sided Fisher’s exact test. d, Schematic of HBR per HLA allele, which is used as input for downstream analyses. e, Summary of the change in comparing individuals with and without EBV DNAemia. P values are the result of a permutation test (n = 100 permutations; two-sided). f, Overview of an inferred model of antigen processing and presentation via MHC, resulting in persistence or clearance of EBV DNA following infection. The images in panels d,f were adapted from ref. ¹⁹, Springer Nature Ltd.