Choosing Controls Wisely to Improve Power in Infectious Disease Genome-Wide Association Studies

Abstract

This thesis addresses a crucial aspect of the design of infectious disease genome-wide association studies (GWAS's): the selection of a control group. In particular, we examined how the level of pathogen exposure within the control group and the level of disease susceptibility in the studied population impacts GWAS power to detect genotype-phenotype associations. We developed a simulation that models an infectious disease GWAS and provides power estimates given control exposure, disease susceptibility, and sample sizes. Simulation data demonstrated that control exposure significantly impacts study power and that ideal GWAS designs will have low disease susceptibility and high control exposure. We also found that GWAS's studying highly susceptible populations are particularly vulnerable to power reductions caused by low control exposure, and recommendations were made for the designs of prospective GWAS's using these results. This research presents a general framework for evaluating GWAS power with respect to control exposure and disease susceptibility, as well as a tool that provides quantitative power estimates to help researchers select an effective control cohort and improve their overall GWAS designs.