PROJECT SUMMARY/ABSTRACT Significance: Novel genetic tools are an emerging avenue for understanding causal relationships between biomarkers and diseases, especially through efficient study designs, such as the genetic instrumental variable method Mendelian Randomization (MR). Tools such as polygenic risk scores (PRS) can be used to efficiently sum the genetic liability of traits and estimate causality in MR. However, both PRS and MR have historically performed poorly in diverse ancestry populations due to their reliance on primarily European ancestry data. This is further compounded by the disproportionate disease burden in marginalized populations due to structural and social factors. We will use the observed association between low density lipoprotein cholesterol (LDL) and coronary artery disease (CAD) from observational epidemiological studies to interrogate the use of a PRS for LDL (PRSLDL) as a valid genetic instrument in MR of diverse groups, modeling genetic and non- genetic disease contributors, and minimizing potential shortcomings of the causal approach. Specific Aims: We aim to 1) characterize differences in the predictive ability of a multi-ethnic PRSLDL for LDL between racial/ethnic groups, with and without adjustment for relevant non-genetic risk factors, 2) determine the effects of individual- and group-level environmental exposures on the PRSLDL performance, stratified by racial/ethnic groups, and 3) assess the causal effect of LDL on CAD in diverse populations, using the PRSLDL as an instrumental variable and stratifying by self-identified racial/ethnic groups. We will use the All of Us Research Program and Population Architecture Using Genomics and Epidemiology (PAGE) cohorts. The long- term objective of this work is to extend the application of new genetic tools to diverse groups. Approach: All analyses will be stratified by racial/ethnic groups to be consistent with prior literature of the relationship between LDL and CAD, particularly with respect to environmental influences. Aim 1 will employ linear regressions of PRSLDL on LDL, with and without adjustment for disease-predictive variables to understand if clinical risk factors change the performance of PRSLDL between racial/ethnic groups. Aim 2 will incorporate regression interaction terms of smoking and air quality with the PRSLDL to inform the differential role of environment in PRSLDL performance by racial/ethnic group. Aim 3 will execute MR, conduct MR sensitivity analyses, and note key MR violations when applied to diverse populations. Fellowship Goals: The proposed study, which addresses several of NHGRI’s strategic goals, will serve as the doctoral dissertation for Ms. Jayati Sharma, a PhD student in the Department of Epidemiology at the Johns Hopkins Bloomberg School of Public Health. A comprehensive training program, including mentored research training from experts in genetic epidemiology, statistical genetics, and cardiovascular genetics, will provide the applican...