Nonlinear Estimation Methods for Mendelian Randomization in Genetic Studies

Mendelian Randomization is a very popular method to estimate causal effects in genomic studies. Recently, there has been much research in instrumental variable (IV) selection. However, these works rely on a linear model framework. In this paper, we focus on IV selection and adopt nonparametric variable selection approaches. We compare linear and nonlinear instrumental variable selection methods. Simulation studies demonstrate the good performance of our methods. We apply our methods for Mendelian Randomization to the UKBiobank data.

Automated summary

Mendelian Randomization is a popular method used in genomic studies to estimate causal effects. This paper focuses on IV selection and introduces nonparametric variable selection approaches, comparing linear and nonlinear instrumental variable selection methods. Simulation studies demonstrate the good performance of the proposed methods, which are also applied to the UKBiobank data for Mendelian Randomization.