🚀 flanders_nf_coloc: Scalable Colocalization with Nextflow

This pipeline wraps the flanders R package to perform fast and scalable colocalization analysis on precomputed AnnData objects.

---

🧬 Step-by-Step Guide

1. Start from an existing AnnData object (produced by flanders::finemap2anndata() or nf-flanders nextflow pipeline)

conda activate /ssu/bsssu/conda_envs_bsssu/flanders_r

library(zellkonverter)
library(SingleCellExperiment)
library(flanders)

# Load your AnnData
ad <- readH5AD("/path/bla/my_anndata.h5ad", reader = "R")

# Generate coloc input table
coloc_input <- anndata2coloc_input(ad)

# Save it to a file
data.table::fwrite(coloc_input, "/path/bla/my_coloc_guide.csv")

Now you have:

• ✅ /path/bla/my_anndata.h5ad
• ✅ /path/bla/my_coloc_guide.csv

These are the two required inputs for the Nextflow pipeline.

---

2. Clone the pipeline

git clone git@github.com:Biostatistics-Unit-HT/flanders_nf_coloc.git
cd flanders_nf_coloc

---

3. Modify the submission script

Edit hcoloc_anndata_example.sh by specifying --annData and coloc_pairwise_guide_table paths to reflect your input paths and desired ID:

#!/bin/bash
#SBATCH --job-name nf-hcoloc
#SBATCH --output nf-hcoloc_%A.log
#SBATCH --partition cpuq
#SBATCH --cpus-per-task 1
#SBATCH --mem 8G
#SBATCH --time 20-00:00:00

module load nextflow/23.10.0

nextflow run flanders_nf_coloc/main.nf \
    --annData "/path/bla/my_anndata.h5ad" \
    --coloc_pairwise_guide_table "/path/bla/my_coloc_guide.csv" \
    --coloc_id my_output_id \
    -resume

---

4. Submit the job

sbatch flanders_nf_coloc/hcoloc_anndata_example.sh

This will produce:

• results/my_output_id_coloc.tsv with all pairwise colocalization results.

---

💡 Notes

For more information, see the upstream flanders package.