PMR(Probabilistic two sample mendelian randomization),is an R package for efficient statistical inference of two-sample MR analysis. It can account for the correlatded instruments and the horizontal pleiotropy, and can provide the accurate estimates of both causal effect and horizontal pleiotropy effect as well as the two corresponding p values.
It is easy to install the development version of PMR package using the 'devtools' package. The typical install time on a "normal" desktop computer is less than one minute.
# install.packages("devtools")
library(devtools)
install_github("yuanzhongshang/PMR")
There are two main functions in PMR package, one is PMR_individual for individual level data, the other is PMR_summary_Egger for summary data. For PMR_indvidual, two pleiotropy model assumptions have been designed, one is the Egger assumption, the other is polygenic assumption (variance component model). Note that the current version of pleiotropy assumption for summary data is Egger, which has been implemented by function PMR_summary_Egger. You can find the instructions by '?PMR_individual' and '?PMR_summary_Egger'.
library(PMR)
?PMR_individual
?PMR_summary_Egger
One simple example to use the package can be found at https://github.com/yuanzhongshang/PMR/tree/master/example
All results from all methods used in the PMR paper can be reproduced at https://github.com/yuanzhongshang/PMRreproduce
This R package is developed by Zhongshang Yuan and Xiang Zhou.