Install grur and its optional engines

2026-08-28

Shared computer setup

See the tgbase genomics computer-setup guide for R, Bioconductor, Miniforge, Conda/Bioconda, executable discovery, reproducibility, and troubleshooting.

Install grur

install.packages(c("BiocManager", "remotes"))
BiocManager::install(c("gdsfmt", "Rsamtools", "SeqArray"))
remotes::install_github("thierrygosselin/grur")

Optional imputation engines

Install only the engine selected with grur_imputations():

install.packages(c(
  "xgboost",
  "ranger",
  "missRanger",
  "randomForestSRC"
))

imputation.method	Package	Role
"xgboost"	xgboost	Gradient-boosted trees
"rf"	randomForestSRC	On-the-fly random-forest imputation
"rf_pred"	ranger	Predictive random forests
"rf_pred" with predictive mean matching	missRanger	Predictive mean matching

For imputation.method = "lightgbm", follow the current LightGBM R installation guide.

Simulation dependencies

simulate_rad() can use rmetasim and fastsimcoal2. Neither is required for missing-data visualization or imputation.

remotes::install_github("stranda/rmetasim")

Obtain fastsimcoal2 from its official website and supply its executable name or full path with fsc.exec. Keep the executable in the shared genomics Conda environment rather than modifying a system directory.

Verify before analysis

library(grur)
packageVersion("grur")
packageVersion("xgboost")  # when using the XGBoost engine