Inferring population histories for ancient genomes using genome-wide genealogies

Abstract

13 Ancient genomes anchor genealogies in directly observed historical genetic variation, and contextualise 14 ancestral lineages with archaeological insights into their geography and lifestyles. We introduce an extension 15 of the Relate algorithm to incorporate ancient genomes and reconstruct the joint genealogies of 14 previously 16 published high-coverage ancients and 278 present-day individuals of the Simons Genome Diversity Project. 17 As the majority of ancient genomes are of lower coverage and cannot be directly built into genealogies, we 18 additionally present a fast and scalable method, Colate, for inferring coalescence rates between low-coverage 19 genomes without requiring phasing or imputation. Our method leverages sharing patterns of mutations dated 20 using a genealogy to construct a likelihood, which is maximised using an expectation-maximisation algorithm. 21 We apply Colate to 430 ancient human shotgun genomes of >0.5x mean coverage. Using Relate and Colate, we 22 characterise dynamic population structure, such as repeated partial population replacements in Ireland, and 23 gene-flow between early farmer and European hunter-gatherer groups. We further show that the previously 24 reported increase in the TCC/TTC mutation rate, which is strongest in West Eurasians among present-day 25 people, was already widespread across West Eurasia in the Late Glacial Period ~10k – 15k years ago, is 26 strongest in Neolithic and Anatolian farmers, and is remarkably well predicted by the coalescence rates 27 between other genomes and a 10,000-year-old Anatolian individual. This suggests that the driver of this signal 28 originated in ancestors of ancient Anatolia >14k years ago, but was already absent by the Mesolithic and may 29 indicate a genetic link between the Near East and European hunter-gatherer groups in the Late Paleolithic. 30 1 bioRxiv preprint doi: https://doi.org/10.1101/2021.02.17.431573; this version posted February 17, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under a CC-BY 4.0 International license. 31