MR Dictionary

Iterative Mendelian Randomization and Pleiotropy (IMRP)

MR method for two-sample MR settings that iteratively searches for horizontally pleiotropic genetic variants being used as instrumental variables (IVs) and calculates the causal effect estimate excluding invalid IVs.

The IMRP method first calculates the MR causal effect estimate using the MR-Egger method. The method then calculates a "pleiotropy test statistic", which is a function of the residual distribution of the causal effect estimation and follows a standard normal distribution if none of the IVs are invalid. A non-zero test statistic for a genetic variant indicates that this may be pleiotropic and, thus, such genetic variants are excluded. The causal effect is then estimated again along with the "pleiotropy test statistic" for identifying any remaining invalid IVs. These steps are repeated iteratively until there is no change in the detected pleiotropic effects and causal effect estimate.


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