TwoStepCisMR: A novel method and R package for attenuating bias in cis-MR

Mendelian randomisation (MR) is an increasingly popular method for strengthening causal inference in epidemiological studies. cis-MR in particular uses genetic variants in the gene region for a genetic proxy of a drug target to provide quasi-experimental evidence for drug efficacy. A major problem for this framework is when the causal variant is correlated to another variant which effects the outcome of interest (confounding through linkage disequilibrium). Methods for correcting bias such as multivariable MR struggle in a cis setting because of the high correlation among genetic variants. Here, we therefore present an alternative method for attenuating bias which does not suffer from this problem. We have additionally developed a simple R package to facilitate the implementation of the method.