Population genomic insights into recent human evolutionary history

Abstract

The advent of large-scale population genomic datasets has enabled detailed inferences regarding human evolutionary history. Demographic changes and positive selection have left their marks on the genome and we can now begin to decipher them. In this dissertation, I present the work I have completed on the topic of human population genomic inference. In chapter 1, I begin by reviewing the importance of human genetic variation and the factors that influence it, focusing on the effects of demographic changes and positive selection. Chapter 2 describes an analysis of genetic ancestry in a worldwide sample of human populations. I show that mitochondrial lineage tests overlook large amounts of variation in genetic ancestry. In chapter 3, I focus on inferences regarding the effective sex ratio in the recent evolutionary past. I present a reanalysis of SNP and resequencing data that resolves a set of conflicting results from previous studies. Using coalescent simulations, I present a model of a recent male bias in effective population size, coupled with an earlier female bias, which is consistent with existing genetic variation on the X chromosome and the autosomes. In chapter 4, I present a comprehensive study of the performance of a battery of neutrality test statistics under a wide range of realistic models of positive selection in recent human evolution. I demonstrate that existing tests perform better than expected for detecting the signatures of a soft sweep from standing variation. Then, I develop a genome-wide approach, the Cumulative Selection Score (CSS), for combining the signals from multiple neutrality test statistics to detect the signatures of positive selection with greater accuracy. By implementing this approach in genomic variation data for chromosome 2, I show that the CSS can be applied to whole-genome datasets. I conclude in chapter 5 by discussing the potential of population genomic inferences and the future of the field.