Inferring the population history of ancient hominins through use of the ancestral recombination graph

This dissertation was inspired by two exciting developments in the field of genomics. The first is the high-quality genome sequencing of ancient archaic individuals, including two Neanderthals and one Denisovan, which has made possible many new insights about human and archaic hominin evolution over the past half million years. Previous studies have demonstrated strong evidence for multiple interbreeding events between these groups, as well as with other unsequenced hominins. The second development is a new method, called ARGweaver, which infers ancestral recombination graphs (ARGs) from the genome sequences of multiple individuals. The ARG describes the genetic relationships between these individuals along the genome, in the form of local trees with branch lengths describing times to the most recent common ancestor. In the first chapter, I provide an introduction to ARGweaver and describe several new features that make it applicable to a wider range of data, including integration over phase, accounting for ancient sampling dates, correcting for low-quality genomes, and sampling under the more accurate SMC' model. In the second chapter, I show how ARGweaver was used to provide strong evidence in favor of a migration event from ancient humans out of Africa over a hundred thousand years ago. These humans likely encountered Neanderthals and admixed with them, leaving segments of their DNA in the Neanderthal genome. In the final chapter, I introduce an extended version of ARGweaver that can sample ARGs conditional on a generic demographic model that may include population divergences and migrations. Once ARGs are inferred under this model, the posterior probability of introgression can be computed along the genome for any migration event. I apply this method to human and archaic hominins, and classify 3% of the Neanderthal genome as potentially introgressed from humans. The properties of these segments suggest that this admixture occurred roughly 250 thousand years ago, and there are no signs that natural selection acted against these regions. I also detect lower levels of introgression from an unknown archaic hominin in the Denisovan genome, and possible traces of the same type of introgression in the Neanderthal genome