LDpred2-pipeline

LDpred2 Docker setup for FinnGen, incorporating the basic dependencies for running LDpred2. Based on the implementation made available at comorment/containers

Docker

Building

In order to build the container locally as ldpred2, issue:

docker build -t ldpred2 -f docker/Dockerfile .

Run

Interactive R session:

docker run -it ldpred2 R

Invoking Rscript on R file:

docker run -it ldpred2 Rscript <filename>.R

Test

Basic test that the built container can be used:

pip3 install pytest  # install the py.test software. Run once.
py.test -v test/test_ldpred2.py

Note: this will only work in environments where Docker does not require sudo.

Reference data

In order to run these LDpred2-based codes on real-world data, the files made available in https://github.com/comorment/ldpred2_ref must be downloaded and found locally.

Scripts

See scripts/README.md for further details.

WDL

The wdl directory contains WDL workflow language file (openwdl.org) templates for converting Plink (.bed, .bim, .fam) files to a LDpred2 compatible format and running LDpred2 itself. Files:

• wdl/convert_bed.wdl: converts genotype files to .rds/.bk files
• wdl/run_ldpred2.wdl: run LDpred2.

Notes

For all workflow files, make sure that all file paths (File) points to correct files. For the LDpred2 part, make sure that the String col_* = "<column>" arguments points to valid columns in the sumstats file. If certain columns like chromosome (col_chr) and position (col_bp) do not exist in the sumstats file, they may be included using the [scripts/complementSumstats.R](scripts/complementSumstats.R) script in R. Note that it requires an external reference file argument (--reference /path/to/HRC.r1-1.GRCh37.wgs.mac5.sites.tab.gz``) from comorment/containers.