downstreamGWAS

downstreamGWAS automates downstream analysis of GWAS summary statistics on HPC systems. Given the output of tidyGWAS, it generates ready-to-run bash scripts for methods like SBayesRC, SBayesS, LD-clumping, and more — handling input conversion, container execution, and SLURM scheduling.

How it works

1. Clean your summary statistics with tidyGWAS::tidyGWAS()
2. Point a pipeline_*() function at the output directory
3. Get a self-contained bash script that converts the data, runs the method inside an Apptainer container, and cleans up

All external software (PLINK, GCTB, LDSC, SBayesRC) runs inside containers — nothing needs to be installed on your cluster beyond Apptainer itself.

Installation

remotes::install_github("ararder/downstreamGWAS")

Quick start

library(downstreamGWAS)

# One-time setup: tell downstreamGWAS where reference data and containers live
setup_dsg(
  storage_root = "/projappl/my_project/downstreamGWAS",
  container_dependency = "ml apptainer"
)

# Generate and submit an SBayesRC job
pipeline_sbayesrc(
  parent_dir = "/path/to/tidyGWAS_output",
  execute = TRUE,
  prepare_inputs = TRUE,
  schedule = schedule_slurm(
    account = "my-project-id",
    mem = "100gb",
    partition = "shared",
    cpus_per_task = 16
  )
)

Available pipelines

Function	Method	Container
pipeline_clumping()	LD clumping (PLINK) + locus merging (bedtools)	genetics_latest.sif
pipeline_sbayesrc()	SBayesRC polygenic scoring	sbayesrc_latest.sif
pipeline_sbayess()	SBayesS (selection/polygenicity)	genetics_latest.sif

See the getting started guide for full documentation.