MR Dictionary

MR with regularization

MR method developed for two-sample MR settings that uses regularization to identify variables that need to be accounted for in the MR model to produce a causal effect estimate robust to violations of MR assumptions.

The method can be used where the number of potentially pleiotropic variables (i.e., the pathways by which the genetic variants being used as instrumental variables (IVs) influence the outcome independent from the exposure) is less than the total number of IVs minus one. Once the pleiotropic variables are identified and excluded, multivariable MR analysis methods can be applied with the remaining variables included in the model to estimate the causal effect. The method can be used with both individual-level (i.e., one-sample MR and summary-level data (i.e., two-sample MR) and does not require any of the genetic variants to be valid IVs.

References

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