Statistical methods for genetic association studies with response-selective sampling designs

Abstract

This dissertation describes new statistical methods designed to improve the power of genetic association studies. Of particular interest are studies with a response-selective sampling design, i.e. case-control studies of unrelated individuals and case-control studies of family members. The statistical methods presented in this dissertation (a) take advantage of information available in the distribution of the covariates in case-control studies by modeling the ascertainment process; (b) incorporate information from both family-based studies and case-control studies of unrelated individuals; (c) use "richer" models of the relationship between genetic variants and phenotypes, compared to models used in standard genetic association studies; and (d) integrate different types of data, such as genomic, epigenomic, transcriptomic and environmental information. Together, these methods will improve the ability of the genetics community to identify the genetic basis of complex human phenotypes.