Additional associations were found with Type 1 Diabetes (T1D), Inflammatory Bowel Disease (IBD) and hypothyroidism, some of which (RA & T1D) even got further support from Mendelian Randomization.

Aggregating the common variants into risk scores, we found that a genetically-predicted EBVread+ based on MHC-I HLA alleles was associated with higher risk of Multiple Sclerosis (driven by HLA-A*02:01), while that of MHC-II HLA alleles was associated with Rheumatoid Arthritis (RA).

Turning to genetics, a GWAS for EBVread+ in UKB identified striking associations at the MHC region, plus 27 loci outside of MHC, most of which were replicated in AoU. Genes at associated loci include ERAP2, CTLA4, CD70.

Exploring which non-genetic factors determine EBVread+, we found strong associations with HIV infection & immunosuppression. When affected individuals were excluded, additional contributors were current (but not former) smoking, male sex, age, GS yield and, interestingly, seasonal time of sampling.

We show, by simulations and additional analyses, that the individuals we capture as EBVread+ are those harboring large amounts of EBV-DNA, indicating insufficient EBV control

We detected EBVreads in 16.2% of 486,315 UKB participants, and this measure (=EBVread+) showed a high specificity when compared to serology data that is available of a subset of individuals. Similar read distributions were obtained in the analysis of 336,123 blood-based GS in All of Us (AoU).

EBV persists lifelong in B cells - allowing it to be captured as part of genome sequencing (GS) of human blood. Such reads are usually discarded, but we extracted them for all individuals of UK Biobank (UKB). During extensive QC, we identified and removed ca. 50 library plates contaminated with EBV