Skip to content

Latest commit

 

History

History
128 lines (99 loc) · 4.23 KB

README.md

File metadata and controls

128 lines (99 loc) · 4.23 KB

SOPA Pipeline Setup Guide

This guide explains how to set up and run the SOPA (Spatial Omics Processing and Analysis) pipeline on the Minerva HPC cluster.

Prerequisites

  1. Access to Minerva HPC cluster with valid credentials
  2. Filtered transcripts must be prepared before running this pipeline
  3. Read or Write access to the following paths:
    • /sc/arion/projects/untreatedIBD/cache/tools/sopa
    • /sc/arion/projects/untreatedIBD/cache/nfs-data-registries/xenium-registry
    • /sc/arion/projects/untreatedIBD/cache/nfs-data-registries/sopa

Environment Setup

1. Set up prep-sopa Environment

# Create conda environment for prep-sopa
cd /sc/arion/projects/untreatedIBD/cache/tools/prep-sopa
conda env create -f environment.yml

2. Set up SOPA and Snakemake Environments

# Create main SOPA conda environment
conda create --name sopa2 python=3.10
conda activate sopa2
cd /sc/arion/projects/untreatedIBD/cache/tools/sopa
pip install -e '.[cellpose,baysor,tangram,wsi]'

# Install Snakemake
conda install -c conda-forge -c bioconda snakemake
pip uninstall numpy
pip install numpy==2.0.2
conda deactivate

3. Set up Baysor executable

# Pull compiled Baysor v0.7.1
wget https://github.com/kharchenkolab/Baysor/releases/download/v0.7.1/baysor-x86_x64-linux-v0.7.1_build.zip
unzip baysor-x86_x64-linux-v0.7.1_build.zip

# Push to Julia config path so Baysor bin found at ~/.julia/bin/baysor
mkdir ~/.julia
cp -r bin/baysor/* ~/.julia

Directory Structure

The SOPA configuration and setup tools are located at: /sc/arion/projects/untreatedIBD/cache/tools/prep-sopa

Setup Steps

  1. Prepare Your Data

    • Ensure your Xenium data is in the correct location in an oba-outputs dir
    • Verify that filtered transcripts are available in the qv20-filtered-transcripts directory

  2. Configuration Setup

    • Copy the example config: example_xenium-multichannel-config.yaml
    • Modify parameters according to your needs
    • Place the config file in your xenium run directory (e.g., /sc/arion/projects/untreatedIBD/cache/nfs-data-registries/xenium-registry/oba-outputs/CHO-001/)

  3. Generate LSF Scripts Activate the prep-sopa environment and run:

    conda activate prep-sopa

python /sc/arion/projects/untreatedIBD/cache/tools/prep-sopa/prep-sopa-run/main.py \
  --id <panel_id>/<run_id> \
  --config-name <your_config.yaml> \
  --project-dir <project_directory>

    Required arguments:

    • --id: Panel/Run ID (e.g., TUQ97N/CHO-001)
    • --config-name: Name of SOPA config file (e.g., cellpose-baysor.yaml)
    • --project-dir: Project directory name for output organization

    Optional arguments:

    • --sopa-source: Path to SOPA workflow directory
    • --conda-env: Path to SOPA conda environment
    • --email: Email for job notifications
    • --queue: LSF queue name
    • --max_ram: Max RAM across all processes

  4. Submit Jobs

    • LSF scripts will be generated in the lsf_scripts directory
    • Submit jobs using: 
      cd oba-outputs/<run_id>/lsf_scripts
bsub < submit_<sample_name>.lsf

Output Structure

All pipeline outputs will be stored in: /sc/arion/projects/untreatedIBD/cache/nfs-data-registries/xenium-registry/sopa/<project_dir>

The pipeline creates:

  1. A timestamped run directory under your project directory
  2. Explorer and Zarr directories for results
  3. Parameter logs and configuration backups
  4. Standard output and error logs

Troubleshooting

  1. If jobs fail, check:

    • Error logs are under the working dir in scratch
      • master log is in workflow/.snakemake/log (e.g. /sc/arion/scratch/tastac01/sopa_50006A-TUQ97N-EA/workflow/.snakemake/log)
      • individual job logs are in workflow/.snakemake/lsf_logs (e.g. /sc/arion/scratch/tastac01/sopa_50006A-TUQ97N-EA/workflow/logs)
    • LSF error files (*.stderr)

  2. Common issues:

    • Missing or incorrect transcript files
    • Insufficient permissions
    • Resource limitations
    • Conda environment activation failures
    • Missing dependencies in environment setup

Notes

  • The pipeline automatically handles transcript file management
  • Results are stored in both scratch and project directories
  • Configuration files must be placed in the specific run directory