Statistical methods for Mendelian randomization in genome-wide association studies: A review

Abstract

Genome-wide association studies have yielded thousands of associations for many common diseases and disease-related complex traits. The identified associations made it possible to identify the causal risk factors underlying diseases and investigate the causal relationships among complex traits through Mendelian randomization. Mendelian randomization is a form of instrumental variable analysis that uses SNP associations from genome-wide association studies as instruments to study and uncover causal relationships between complex traits. By leveraging SNP genotypes as instrumental variables, or proxies, for the exposure complex trait, investigators can tease out causal effects from observational data, provided that necessary assumptions are satisfied. We discuss below the development of Mendelian randomization methods in parallel with the growth of genome-wide association studies. We argue that the recent availability of GWAS summary statistics for diverse complex traits has motivated new Mendelian randomization methods with relaxed causality assumptions and that this area continues to offer opportunities for robust biological discoveries.

Keywords: Causal inference; Confounding; Genome-wide association study; Genomics; Horizontal pleiotropy; Mendelian randomization.

Graphical abstract

Fig. 1

Upward trend in article counts by year for Google Scholar keyword searches: 1. Mendelian randomization and 2. Genome-wide association study.

Fig. 2

Two-sample MR with GWAS summary statistics methods publications density increases over time.

Fig. 3

Causal diagrams for MR. A. Scenario where the genetic variant affects an intermediate variable on the pathway to the exposure. Because the intermediate affects the outcome through a pathway that doesn’t involve exposure, this scenario violates the exclusion restriction assumption. B. Scenario where the genetic variant affects an exposure, which in turn affects an outcome, possibly in the presence of unmeasured confounding. C. Correlated horizontal pleiotropy occurs when the genetic variant affects the exposure, which in turn affects the outcome, and the genetic variant affects the unobserved confounder, which in turn affects the exposure and the outcome independently.

Fig. 4

Decision tree diagram for choosing among multiple-SNP MR methods.