Recent advances in large-scale whole exome sequencing and a method for region-based estimation of rare variant effect sizes enable more efficient quantification of their magnitude and direction. Building on this framework, we propose a gene-based caus...
Recent advances in large-scale whole exome sequencing and a method for region-based estimation of rare variant effect sizes enable more efficient quantification of their magnitude and direction. Building on this framework, we propose a gene-based causal inference approach that integrates gene-level rare variant scores into Mendelian randomization (MR). In this model, composite genetic scores derived from rare variant effect sizes serve as instrumental variables, allowing estimation of causal effects between exposures and outcomes through a two-stage least squares framework. We validated the method using simulations under varying confounding structures to assess power and type I error, and further applied it to UK Biobank data. Collectively, our findings are expected to demonstrate that incorporating rare variants into MR broadens its applicability and enhances the resolution of causal inference for complex traits and diseases.