Statistical Methods for the Analysis of Autosomal and X Chromosome Genetic Data in Samples with Unknown Structure

Abstract

Genome-wide association studies (GWAS) and sequencing association studies are routinely conducted for the mapping of genes to complex traits. Genetic variants on the X chromosome could potentially play an important role in some complex traits, however, statistical methods for association studies have primarily been developed for variants on the autosomal chromosomes with significantly less attention given to the X chromosome. Existing association methods for variants on the autosomal chromosomes will typically not be valid for the analysis of X-linked variants due to the X chromosome having a different correlation structure than the autosomes as well as copy number differences for males and females on the X. This dissertation develops and applies new statistical methodology for genetic analysis of variants on the X chromosome. In particular, we focus on methods that are computationally feasible for large-scale genomic data for detecting genetic associations with common and rare variants from GWAS and sequencing studies. Furthermore, the proposed methods allow for valid genetic analysis in the presence of complex sample structures, such as population structure and cryptic relatedness among sampled individuals. Another aspect of this dissertation is the development of statistical methods for inference of heterogeneity in ancestry across the genome (including the X chromosome) in recently admixed populations, such as African Americans and Hispanics, who have experienced admixing within the past few hundred years from two or more continental groups that were previously isolated.