Inference of Recent Demographic History of Population Isolates using Genome-wide High Density SNP Arrays and Whole Genome Sequences

Abstract

In this dissertation I addressed the problem of SNP array bias when finding runs of homozygosity. I demonstrated the pitfalls of using uninformed methods for finding runs of homozygosity and provide better alternatives, including a more reliable algorithm for identifying runs of homozygosity than the most commonly used program. I then provide a review of Ashkenazi population genetics. Next, I developed software to efficiently run millions of whole chromosome simulations, which is publicly available through GitHub, DockerHub, and on the CyVerse Discovery Environment. I applied my computational method to use Approximate Bayesian Computation to test models of Ashkenazi Jewish demographic history. I found that the Ashkenazi Jews are comprised of genetically distinct subgroups from Eastern and Western Europe, as a result of massive population growth in the Eastern Ashkenazi Jews, but not in the Western Ashkenazi Jews. I further confirmed that the Ashkenazi Jews do not primarily originate from Khazaria. Finally, I created a correction for SNP array ascertainment bias in the median and total length of runs of homozygosity, and applied this correction to world-wide human populations. However, I found that ascertainment bias plays a minor role compared to SNP array bias in human populations.