Package: bnmr2
Type: Package
Title: A package to conduct Bayesian Network-based Mendelian randomization using individual-level GWAS data
Version: 0.3.1
Author: Jianle Sun
Maintainer: The package maintainer <sjl-2017@sjtu.edu.cn>
Description: An R package to conduct causal estimation between exposure and outcome on GWAS data using the Bayesian Network-based Mendelian Randomization (BNMR), which is a two-stage framework to deal with imperfect genetic instruments (horizontal pleiotropy, linkage disequilibrium, epistasis, etc.). It selects variants with a direct effect on the exposure as instrumental variables via random graph forest (RGF), an ensemble approach comprised of a series of Bayesian network structure learning processes within sampled variants and exposure. It then uses the Bayesian Mendelian randomization (BMR) with a shrinkage prior on the nuisance parameters to cope with potential horizontal pleiotropy and obtain a robust estimate. 
License: GPL
Encoding: UTF-8
LazyData: False
Imports:
    bnlearn,
    rstan,
    MendelianRandomization,
    plyr,
    dplyr,
    AER,
    ivmodel,
    parallel,
    purrr
RoxygenNote: 7.2.3
