Estimating and visualising multivariable Mendelian randomization analyses within a radial framework.

ABSTRACT:

Background

Mendelian randomization (MR) is a statistical approach using genetic variants as instrumental variables to estimate causal effects of a single exposure on an outcome. Multivariable MR (MVMR) extends this to estimate the direct effect of multiple exposures simulatiously. MR and MVMR can be biased by the presence of pleiotropic genetic variants in the set used as instrumental variables, violating one of the core IV assumptions. Genetic variants that give outlying estimates are often considered to be potentially pleiotropic variants. Radial plots can be used in MR to help identify these variants. Analogous plots for MVMR have so far been unavailable due to the multidimensional nature of the analysis.

Methods

We propose a radial formulation of MVMR, and an adapted Gal

SUBMITTER: Spiller W 

PROVIDER: S-EPMC11684766 | biostudies-literature | 2024 Dec

REPOSITORIES: biostudies-literature