Using Multi-Trait Genome-Wide Association Summary Statistics to Understand Genetic Architecture and Pleiotropy: Novel Methods and Translational Applications to Cancer

Abstract

Genetic epidemiology is a subject that studies how genetic factors impact the health and disease status among populations. With the development of modern statistics and data science, statistical and big-data techniques are increasingly used to study genetic epidemiology. In this thesis, I developed multiple statistical methods to tackle the challenges in the

field. These challenges include understanding the genetic architectures of complex traits for a better study design in the future, prioritizing genetic variants and genes that are potentially associated with one trait or multiple traits (i.e., pleiotropic), and exploring the shared biological pathways across multiple genetically correlated traits. These methods are built upon multi-trait genome-wide association summary statistics and leverage the concept of cross-trait meta- analysis. The methods are also computationally efficient and can be applied to a wide range of polygenic phenotypes. I applied these novel methods to study the genetic profiles of twelve cancer types and

have made a series of consequential discoveries including but not limited to novel pleiotropic variants and regions, novel cancer-associated genes, and shared cancer-related biological pathways. Among them, some have already been validated through molecular genetics experiments, and others offer an opportunity for further study in the field of genetic research.