Methods for Mendelian randomization and mediation analysis using integrative genetic and genomic data for breast cancer

Project summary Haoyu Zhang, PhD is a statistician whose ultimate career goal is incorporating advanced causal inference techniques into genetics and biological research in order to make impactful advances in epidemiological and clinical decision-making. The research he proposes will develop powerful causal inference approaches to

identify causal risk factors and underlying genetic pathways leading to the risk of breast cancer. Candidate: Dr. Zhang is a postdoctoral fellow in the Department of Biostatistics at Harvard T.H. Chan School of Public Health (HSPH). He completed a Ph.D. in Biostatistics at Johns Hopkins University. His previous work

in breast cancer genome-wide association studies (GWAS) focusing on identifying genetic associations and building polygenic risk has prepared him to conduct the proposed research. The proposed career development plan will build upon his previous training with three training goals to enhance trajectory toward becoming an

independent investigator: 1) acquire and apply cutting edge causal inference methodologies to apply on large genetic datasets; 2) gain knowledge in molecular biology and cancer; 3) develop leadership and professional skills to conduct multidisciplinary analysis. Mentors/Environment: Dr. Zhang has assembled a strong mentoring committee with complementary

expertise in the required fields for the proposed research. All the mentors have committed to meet with him in a regular basis and participate the advisory meeting to oversight his training and research progress every six months. As an institution, HSPH is committed to help young researchers. Dr. Zhang will have access to

professional and career development resources, which include professional development courses, writing and editing support for papers and grant applications, etc. Research: Risk factors for the breast cancer include reproductive and life events (collectively classic risk factors) and genetic factors; however, the causal associations and pathways linking these risk factors with

breast cancer are unclear. To solve these two issues, He will develop a robust and powerful approach for Mendelian randomization analysis to estimate the causal effects between classic risk factors and breast cancer risk (Aim 1). He will also develop a causal mediation approach integrating functional annotation datasets to

identify the underlying pathways for breast cancer risk (Aim 2). In Aim 3, he will apply both the standard approaches and novel approaches developed in Aim 1 and 2 on the largest breast cancer GWAS dataset from the multi-ethnic Confluence Project. The results of this proposal will provide advanced statistical tools to

identify causal effect, elucidate the underlying genetic pathways and guide developments of personalized therapeutics and prevention strategies. The proposal will also provide him the required training and research experience to become an independent research with casual inference and breast cancer expertise.