BESIDE-MR

BayEsian Set IDEntification Mendelian randomization

Installation

To install BESIDEMR directly from the GitHub repository, use (please ensure you have the devtools package installed already):

library(devtools)
install_github("CYShapland/BESIDEMR")

To update the package just run the install_github("CYShapland/BESIDEMR") command again.

Description

The main function is BESIDE_MR to perform BayEsian Set IDEntification Mendelian randomization (BESIDE-MR). We develop a bespoke Metropolis-Hasting algorithm to perform the search using the recently developed Robust Adjusted Profile Likelihood (MR-RAPS) of Zhao et al as the basis for defining a posterior distribution that efficiently accounts for pleiotropic and weak instrument bias. BESIDE-MR can be extended from a standard one-parameter causal model to a two-parameter model, to allow a large proportion of SNPs to violate the Instrument Strength Independent of Direct Effect (InSIDE) assumption.

BESIDE_MR returns an object of class beside, consists of the posterior of effect estimate, pleiotropy variance and instrument inclusion indicator variable from each iteration. As the estimation of variance is challenging, we have included tau_estimate with the options of DL estimate and Full_Bayes, where the former is a plug-in estimate for pleiotropy variance.

Update

In August 2021, we have added a penalization term ($\eta$) within likelihood function of BESIDE-MR. The parameter $\eta$ dictates the size of models BMA explores the most. Setting a large positive $\eta$, the likelihood will increase with number of instruments, then BMA will favour models with many instruments. And hence for negative $\eta$, BMA will favour models with fewer instruments.

Citation

The corresponding paper can be accessed at:

Shapland, C.Y., et al., Profile-likelihood Bayesian model averaging for two-sample summary data Mendelian randomization in the presence of horizontal pleiotropy. bioRxiv, 2020

License

This project is licensed under GNU GPL v3.