Statistics Department Seminar Series: Jingshu Wang, Assistant Professor, Department of Statistics, University of Chicago

Abstract: Understanding the pathogenic mechanisms of common diseases is a fundamental goal of clinical research. As randomized controlled experiments are not always feasible, researchers are looking towards Mendelian Randomization (MR) as an alternative method for probing the causal mechanisms of common diseases. In MR, natural genetic variations are used as instrumental variables to unbiasedly estimate causal effects in the presence of unmeasured environmental confounding. Partly due to its convenience, MR has recently been widely employed in epidemiology and other related areas of population science. However, the phenomenon that “all genes affect every complex trait” complicates Mendelian Randomization (MR) studies as most genetic variants will likely be invalid instruments. Furthermore, causal relationships between risk factors and diseases can be more complex than depicted in a simple diagram. During the presentation, I will describe a new comprehensive framework that we have developed to address multiple challenges in MR. Specifically, I will focus on the issues of pervasive horizontal pleiotropy, weak genetic instruments, and the temporal relationship between risk factors and disease progression. I will discuss our statistical theories and methodologies, with a few case studies at the end of the talk.