Recent technological advancements have enabled us to collect an unprecedented amount of genetic epidemiological data. The overarching goal of these genetic epidemiology studies is to uncover the underlying biological mechanisms so that improved strategies for disease prevention and management can be developed. To efficiently analyze and interpret high-dimensional biological data, it is imperative to develop novel statistical methods as conventional statistical methods are generally not applicable or are inefficient. In this dissertation, we introduce three novel, powerful and computationally efficient kernel machine set-based association tests for analyzing high-throughput genetic epidemiological data. In the first chapter, we construct a test for identifying common genetic variants that are predictive of a time-to-event outcome. In the second chapter, we develop a test for identifying gene-environment interactions for common genetic variants. In the third chapter, we propose a test for identifying gene-environment interactions for rare genetic variants.
