pairwise-colocalization-smk

Purpose

This pipeline performs pairwise colocalization analysis between genetic signals from different GWAS summary statistics using the coloc R package. It identifies pairs of signals that are close together, extracts the relevant regions from summary statistics, and runs colocalization for each pair.

This pipeline was originally written as a set of array jobs, and is adapted from the original pipeline, which can be viewed under the slurm/ directory.

Overview

The workflow is managed by Snakemake and is modular, allowing for scalable and reproducible analysis of many signal pairs.

Inputs

• Input signals file: CSV file with columns:

  • chromosome: Chromosome number
  • position: Genomic position (base pair)
  • gwas: Name of the GWAS (should match summary stats file prefix)
  • (other columns allowed, but not required)
  • Example: test/data/input_signals.csv

• Summary statistics files: One file per GWAS, named {gwas}.txt and placed in the directory specified in workflow/config.yaml (default: test/data/sumstats/). Each file should be tab-delimited and contain at least:

  • chromosome
  • position
  • rsid
  • beta
  • se

Outputs

• outputs/pairs_to_test.csv: List of all signal pairs to be tested
• outputs/extracted_regions/: Extracted region files for each pair and GWAS
• outputs/coloc_results/: Colocalization results for each pair
• outputs/coloc_credible_sets/: Credible set (if applicable) for each result
• outputs/joined_credible_sets/: For colocalized signals, credible sets are joined

Requirements

• Snakemake (v6+ recommended)
• R (v4+ recommended)
• R packages: coloc, dplyr, readr, tidyr

Usage

1. Configure your input files

   • Place your input signals CSV in test/data/input_signals.csv (or update the path in workflow/config.yaml)
   • Place your summary statistics files in test/data/sumstats/ (or update the path in workflow/config.yaml)
   • If you do not make any changes, then the pipeline will run on the test dataset

2. Edit configuration

   • Edit workflow/config.yaml to set the correct paths and distance threshold (in base pairs)

3. Run the pipeline

   snakemake --cores 4 all

   (Adjust the number of cores as needed)

4. View results

   • Colocalization results for each pair will be in test/outputs/coloc_results/

Pipeline Steps

1. find_pairs: Identifies all pairs of signals within the specified distance on the same chromosome, from different GWAS.
2. extract_regions: For each pair, extracts the relevant region from each GWAS summary stats file.
3. run_coloc: Runs the coloc R package on each pair of extracted regions and outputs the results.