MR Dictionary

Bayesian multi-instrument Mendelian randomization (MIMR)

Bayesian approach for two-sample MR settings that models and accounts for instrumental variables (IVs) that exert horizontally pleiotropic effects on the outcome independent of the exposure.

This Bayesian method represents horizontally pleiotropic effects by unknown parameters in the MR estimation model and imposes a shrinkage prior distribution that assumes an unspecified subset of the effects to be zero. The resulting posterior distribution can then be sampled by Markov chain Monte Carlo methods to obtain the causal effect estimate and related confidence intervals. The method is reasonably robust to the presence of directional pleiotropy and moderate correlation between the IVs. The method can also be extended to estimate the direct and indirect effects of scenarios with two exposures on an outcome using MR methodology.

References

Other terms in 'Pleiotropy-robust two-sample MR methods':