Variant Selection and Aggregation of Genetic Association Studies in Precision Medicine

Precision medicine provides tailored treatments to patients based on their demographic and genomic profiles, and is a promising approach to improving treatment efficiency. Genetic biomarkers identified from association studies are useful in disease risk prediction, subgroup identification and heterogenous drug responses, all which are important steps in precision medicine. Marginal tests of association are powerful for common and relatively major variant identification, but variants with low frequency or modest association can explain additional disease risk and may be important for pharmaceutical intervention. In this chapter, we review set-based association methods with a focus on variant selection and aggregation involving low frequency and signal variants. We present a unifying framework as well as a detailed classification of methods. Also discussed in this article are the control of type I error and replication of discoveries. Simulation studies are conducted to demonstrate the effectiveness of data-driven methods under different assumptions of causal variants. We conclude with some discussions on the challenges and future research directions.