A robust and efficient method for Mendelian randomization with hundreds of genetic variants: unravelling mechanisms linking HDL-cholesterol and coronary heart disease

Abstract

Mendelian randomization (MR) investigations with large numbers of genetic variants are becoming increasingly common. However, the reliability of findings from a MR investigation is dependent on the validity of the genetic variants as instrumental variables. We developed a method to identify groups of genetic variants with similar causal effect estimates, which may represent distinct mechanisms by which the risk factor influences the outcome. Our contamination mixture method is a robust and efficient method for valid MR in the presence of invalid IVs. Compared to other robust methods, our method had the lowest mean squared error across a range of realistic scenarios. The method is fast and efficient, and can perform analysis with hundreds of variants in a fraction of a second. In a MR analysis for highdensity lipoprotein (HDL) cholesterol and coronary heart disease (CHD) risk, the method identified 11 variants associated with increased HDL-cholesterol, decreased triglyceride levels, and decreased CHD risk that had the same directions of associations with platelet distribution width and other blood cell traits, suggesting a shared mechanism linking lipids and CHD risk relating to platelet aggregation.

A Comparison Of Robust Mendelian Randomization Methods Using Summary Data

“…The contamination mixture method assumes that only some of the genetic variants are valid IVs [30]. We construct a likelihood function from the ratio estimates.…”

A Comprehensive Evaluation of Methods for Mendelian Randomization Using Realistic Simulations and an Analysis of 38 Biomarkers for Risk of Type-2 Diabetes

“…Very recently, we proposed the method MRMix which uses an underlying mixture model to implicitly distinguish valid and invalid instruments and estimate the causal effect accordingly. Most recently, a variation of mixture-model based method has been proposed, called contamination mixture 19 , which models the ratio estimates using normal-mixture with prespecified variance parameters.…”

“…Further, many simulation studies also directly simulate data on the instruments ignoring the process that instruments in reality are selected to be SNPs that reach genome-wide significance in an underlying genomewide association study -as a result of which there should be a close relationship between sample size, number of available instruments, their average effect-sizes and precision of their estimated effects. Previous MR studies have used a fixed number of IVs and a fixed sample size 4,13,19 or vary one of them without the other [16][17][18]29,30 , and generate the effects of genetic instruments from a fixed distribution without varying with the sample size or the number of IVs.…”

Bayesian variable selection with a pleiotropic loss function in Mendelian randomization

“…The contamination mixture approach Burgess et al (2019) uses a mixture model for the univariate causal effect estimates. The mixture contains two components: valid instruments are assumed to be normally distributed around the true causal effect and invalid instruments are normally distributed around zero.…”

“…In practical applications, we typically do not know which genetic variants exhibit pleiotropic effects and there is need for methods to perform Mendelian randomization in the presence of pleiotropic variants. Traditional approaches include MR-Egger regression (Bowden et al, 2015) and median estimation , while several algorithms for pleiotropy-robust Mendelian randomization have been developed recently (Hartwig et al, 2017;Kang et al, 2016;Rees et al, 2019;Burgess et al, 2018;Verbanck et al, 2018;Zhao et al, 2018;Qi and Chatterjee, 2018;Bowden et al, 2018;Burgess et al, 2019). Recent reviews and comparison of the various methods can be found in Slob and Burgess (2019) and Qi and Chatterjee (2019).…”