Abstract
Computational HLA typing has surged as a cost-effective strategy to uncover questions regarding the evolution of the HLA system, enabling immunogenic characterization from ancient DNA (aDNA) data. Nevertheless, it remains to be seen whether these methods are suitable for analyzing aDNA generated without target-enrichment. To investigate this, we evaluated the performance of five HLA typing tools using present-day data with simulated profiles typical of aDNA, as well as from high-coverage aDNA genomes downsampled at different read depths. We found that characterization of Class I genes at the first field resolution is feasible at read depths as low as 2x, where it retains an accuracy of ≈ 80%. Next, we used this insight to characterize HLA evolution in Europe from 154 ancient genomes by detecting allele frequency changes throughout distinct prehistoric European populations. We observed important shifts in alleles associated with infectious and autoimmune diseases, most of which are found by contrasting the HLA landscape of Neolithic Farmers to that of present-day. Interestingly, several of these observations are in line with findings that have been previously reported by target-enrichment-based studies. Our results highlight the feasibility of applying HLA typing on shotgun aDNA data to examine the evolution of this loci during important transitions.