I screened ancient skeletons from 23 archaeological sites spanning the Paleolithic, Neolithic, Eneolithic, Bronze Age, Iron Age, Antiquity, and the Middle Ages. DNA extraction was performed using a silica column-based protocol, targeting the densest portions of the petrous bone or teeth to maximize endogenous DNA yield. Libraries were treated with the USER enzyme to mitigate DNA-damage-induced transitions and subsequently screened to assess ancient human DNA content using an Illumina MiSeq platform. We then sequenced 52 of these samples to 1X genome-wide coverage, with at least five samples represented per time period. The sequencing was performed using the NovaSeq 6000 platform. We developed in-solution RNA baits to capture and sequence the MEFV gene at high coverage (located on chromosome 16). This was done for the majority of the screened samples. We processed the data using standard procedures for ancient genomics and rigorous quality control measures. Data authenticity was evaluated by analyzing the average sequence length, identifying patterns of molecular damage, and estimating contamination rates in the mitochondrial and X chromosomes. We also conducted radiocarbon dating to place them precisely along the time scale. In addition to the ancient data generated in this project, I obtained population-level MEFV gene haplotype frequencies from 68 unrelated modern Armenians. I also integrated our data with a recently published study that featured 36 whole-genome sequences (30x coverage) from modern Armenian individuals. In our recent published study (Hovhannisyan et al. 2025), we observed an unexpected finding: nearly half of the individuals in our dataset (16 out of 36) carried pathogenic or likely pathogenic variants in the MEFV gene - a frequency significantly exceeding the 20% reported in previous studies. This result underscores the limitations of conventional diagnostic approaches, such as strip assays, which appear to lack the sensitivity required to accurately identify pathogenic or likely pathogenic mutations in the Armenian population. For the population genetic analyses, we compared Armenians with reference datasets from the broader geographic region, using an array of conventional population genetics models, such as Principal Component Analyses, ADMIXTURE, fineStructure, and various f-statistics. To detect selection, we used a Bayesian statistical framework which accounts for demography.
