README.md
In YangLabHKUST/MRAPSS: What the Package Does (One Line, Title Case)

MRAPSS

The MRAPSS package implement the MR-APSS approach to test for the causal effects between an exposure and a outcome disease.

The MR-APSS is a unified approach to Mendelian Randomization accounting for pleiotropy, sample overlap ans selection bias using genome wide summary statistics.

Specifically, MR-APSS uses a background-foreground model to characterize both SNP-exposure effects and SNP-outcome effects, where the background model accounts for the signals due to the shared heritable factors and the foreground model captures the valid signal for causal inference. Building upon the background-foreground model, MR-APSS further takes into account the issues of selection bias and sample overlapping, making it widely applicable for real data analysis.

Installation

#install.packages("devtools")
devtools::install_github("YangLabHKUST/MRAPSS")

Usage

We illustrate how to analyze GWAS summary level data using the MRAPSS software by an real example, i.e. BMI (UKB) (exposure) and T2D (outcome). The MRAPSS analysis comprises five steps:

Step 1: Download GWAS summary-level data from public resources

Step 2: Format data

Step 3: Harmonise datasets and estimate nuisance parameters

Step 4: IVs selection and LD clumping

Step 5: Fit MRAPSS

Step 6: Visualize

The tutorial: A real example for perfroming GWAS summary-level data based MR analysis with MRAPSS package provides details for each step.

To have a quick look at the MRAPSS, you can skip Steps 1-4 and directly jump to Step 5 and Step 6 to fit MRAPSS using the outputs we have prepared.

library(MRAPSS)
exposure = "BMI"
outcome = "T2D"
Threshold = 5e-05  # The default p-value threshold IV selection 
data(Sigma_err)
data(Omega)
data(MRdat)
MRres = MRAPSS(MRdat,
               exposure="BMI",
               outcome= "T2D",
               Sigma_err = Sigma_err,
               Omega =  Omega ,
               Cor.SelectionBias = T)
MRplot(MRres, exposure="BMI", outcome="T2D")

The "BMI~T2D" example with 1228 IVs takes about 1 minutes tested on MAC OS 10.14.6 with 1.4 GHz Intel Core i5,16 GB 2133 MHz LPDDR3 and R version 3.6.1.

Reproduce

We provide the downlowad links for 20 GWAS summay datasets used in MR-APSS paper which can be found here. The R code for processing datasets and implementing other compared MR methods as well as the results can be downloaded here

Reference

Xianghong Hu, Jia Zhao, Heng Peng, Yang Wang, Xiang Wan, Yang Can, MR-APSS: a unified approach to Mendelian Randomization accounting for pleiotropy, sample overlap ans selection bias using genome wide summary statistics.