MR Dictionary

MR-Clust

A two-sample MR method that identifies distinct clusters of genetic variants being used as instrumental variables (IVs) dependent on their individual causal estimates and how similar they are.

Causal effect estimates of multiple genetic IVs are likely to be different if there are distinct causal mechanisms by which an exposure influences an outcome. For example, the exposure may be a composite trait, comprising multiple components or the exposure may influence the outcome through distinct causal pathways, each of which affect the outcome to a different magnitude. Equally, the causal effect estimates of multiple genetic IVs may be different due to violations of MR assumptions. For example, causal pathways by which the genetic variants influence the outcome, which may represent horizontal pleiotropy independent of the exposure. MR-Clust is an algorithm that identifies such clusters of genetic IVs, accounting for differential uncertainty in the causal estimates and including "null" and "junk" clusters to provide protection against the detection of spurious clusters. This clustering approach may lead to a better biological understanding of the causal relationship between the exposure and outcome, and the possible mechanisms by which this occurs.

References

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