different interventions simultaneously. Genetic methods are useful to infer causality because genetic variants are present from birth and therefore unlikely to be confounded with environmental factors. Mendelian randomization (MR) is an analysis that uses genetic variants, which are expected to be independent of confounding factors, as instrumental variables to test for causality9–11. MR can be used to infer credible causal associations when RCTs are not feasible or as a strategy to rank order candidate causal associations to be prioritized for follow-up in RCTs. MR is becoming increasingly efficient and cost-effective given the ever-growing data curated from recent genome-wide association studies (GWAS). The large amount of GWAS data available in the public domain provide a great opportunity for methods that are able to make inference about causality by integrating summary-level GWAS data from different studies12–16. We have previously shown that the power of an MR analysis could be greatly improved by exploiting GWAS summary data from two independent studies with large sample sizes, and have applied a summary data-based MR (SMR) approach to test whether the effects of genetic variants on a phenotype are mediated by gene expression17.
