SCENIC Snakemake Workflow

A comprehensive Snakemake workflow for running SCENIC analysis on single-cell RNA-seq data.

Overview

This workflow implements the complete SCENIC pipeline using Snakemake, including downstream analysis.

Features

• Data preprocessing: Quality control and filtering of single-cell data
• Network inference: Gene regulatory network inference using GENIE3
• Regulon discovery: Identification of regulons using cisTarget
• Activity scoring: Calculation of regulon activity scores (AUC)
• Visualization: Comprehensive plots and reports
• Scalable: Parallel processing support
• Reproducible: Conda environments and version control

Workflow Steps

1. Expression Matrix Preparation

   • Normalize expression data
   • Log-transform values
   • Format for SCENIC input

2. Gene Regulatory Network Inference

   • Run GENIE3 algorithm
   • Generate gene-gene adjacency matrix

3. Regulon Creation

   • Use cisTarget for motif enrichment
   • Create transcription factor regulons

4. Activity Scoring

   • Calculate AUC scores for each regulon
   • Generate binary activity matrix

5. Visualization & Analysis

   • Regulon activity heatmaps
   • UMAP plots colored by regulon activity
   • Regulon Specificity Score (RSS) analysis

Quick Start

1. Clone and Setup

git clone <repository-url>
cd scenicSnake

2. Install Dependencies

Create the conda environment:

conda env create -f envs/scenic.yaml
conda activate scenic

3. Configure Your Analysis

Edit the configuration files:

• config/config.yaml: Main workflow parameters

4. Prepare Your Data

Create an AnnData object (.h5ad) of the preprocessed data.

5. Run the Workflow

# Dry run to check the workflow
snakemake -n

# Run the complete workflow
snakemake --cores 8 --use-conda

# Run specific steps
snakemake results/scenic/regulons.json --cores 4 --use-conda

Configuration

Main Parameters (config/config.yaml)

Make sure the config.yaml file is updated prior to running.

Output Files

Core Results

Files will be saved for each split condition if applicable.

• results/scenic/adjacencies.tsv: Gene-gene adjacency matrix
• results/scenic/regulons.json: Discovered regulons
• results/scenic/auc_matrix.csv: Regulon activity scores
• results/scenic/binary_regulon_activity.csv: Binary regulon activity

Visualizations

• results/plots/regulon_heatmap.pdf: Regulon activity heatmap
• results/plots/umap_regulon_activity.pdf: UMAP with regulon overlay
• results/plots/rss_plot.pdf: Regulon specificity scores

Reports

• results/reports/scenic_report.html: Comprehensive analysis report

Performance Optimization

1. Use cluster execution:

snakemake --cluster "sbatch --time={resources.time} --mem={resources.mem}" --cores 32

2. Adjust resource allocation in config/config.yaml

Citation

If you use this workflow, please cite:

• SCENIC: Aibar et al. Nature Methods (2017)
• Snakemake: Köster & Rahmann, Bioinformatics (2012)
• scanpy: Wolf et al. Genome Biology (2018)

License

This workflow is released under the MIT License.