Inferring causal effect of risk factors on chronic disease using genetics
The availability of genetic data from large-scale biobanks allows use of genetic instrumental variables to infer causality through Mendelian randomization. This technique circumvents the problem of confounding that is so ubiquitous in observational studies, particularly for questions that cannot be explored through randomized trials. In this talk, I will discuss examples of application of this method to studying cardiovascular and oncologic effects due to two common risk factors: alcohol consumption and age of menopause. I will discuss future directions of discovering and using genetic instrumental variables to understand additional factors related to earlier life events and later life chronic diseases. Finally, I will also discuss future directions in building undergraduate data science research opportunities.