LDSC Pipeline

A simple Snakemake pipeline to run munge_sumstats and LD score regression, including the download of all necessary dependencies.

Prerequisites

Install Snakemake (tested with version 7). This step is described in this repository. Follow these two steps:

1. Install Conda
2. Install Snakemake via Conda

Configuration

Add your own summary statistics to the file resources/traits.tsv. You can find some example records to modify. The columns are used as follows:

• trait_name → Name you would like to assign to the respective trait
• ss_path → Path where your summary statistics are stored
• snp → RSID column name
• a1_effect → Effect allele column name
• a2_non_effect → Non-effect allele column name
• add_munge_args → Extra parameters for merge_sumstats
• no_munge → If set to 1, the munge step will be skipped

Run

Once configured, execute Snakemake within the root folder of the repository:

snakemake -c1 --use-conda
# Increase -c1 to e.g. -c4 to use more (here: 4) cores

Heritabilities of individual traits and genetic correlations between all possible pairs of traits are calculated and stored in the results/ldsc/ folder.

Depiction

Genetic correlations can be visualized using the script in the notebooks/ folder. To generate the plots:

1. Run the bash commands given in gather_genetic_correlations.sh
2. Run the R script display_genetic_correlation.R

Acknowledgment

The Snakemake workflow is based on the rules from this repository.