scRNA-seq Snakemake Pipeline

This repository contains a fully reproducible pipeline for preprocessing, quality control, doublet detection, and basic downstream analysis of single-cell RNA-seq data using Snakemake and Docker.

---

🔧 Features

• Environment-controlled via Docker
• Flexible Snakemake workflow driven by a metadata.csv
• Species-aware processing (human or mouse)
• FASTQ → Cell Ranger → CellBender → scDblFinder → Seurat QC
• Optional output as Seurat, SingleCellExperiment, or AnnData
• Clustering and marker gene identification

---

📁 Directory Structure

├── config/
│   ├── config.yaml           # Pipeline configuration
│   └── metadata.csv          # Sample metadata (sample ID, species, fastq path)
├── data/                     # FASTQ files per sample
├── results/                  # All pipeline outputs
│   ├── cellranger/
│   ├── cellbender/
│   ├── fastqc/
│   ├── r_analysis/
│   └── scdblfinder/
├── scripts/                  # R scripts
├── Snakefile                 # Main Snakemake pipeline
├── Dockerfile                # Environment definition

---

📥 Requirements

• Docker
• Snakemake (or run inside the container)
• FASTQ files named: sampleID_R1.fastq.gz, sampleID_R2.fastq.gz

---

🧾 Input Files

1. metadata.csv

sample,species,fastq_path
sample1,human,data/fastq/sample1
sample2,mouse,data/fastq/sample2

• species: either human or mouse
• fastq_path: path to folder with paired-end FASTQ files

2. config.yaml

genomes:
  human: /opt/refdata-gex-GRCh38-2020-A
  mouse: /opt/refdata-gex-mm10-2020-A

paths:
  metadata: config/metadata.csv
  output_dir: results
  r_script: scripts/scRNA_qc_plots.R

---

🚀 Running the Pipeline

Option A: Inside Docker

1. Build the image:

docker build -t scrna_pipeline .

2. Run the container:

docker run --rm -v $PWD:/workspace -w /workspace -e COMBINED_OUTPUT_TYPE=sce scrna_pipeline snakemake --cores 8

Option B: Local (outside Docker)

Ensure R, Snakemake, Cell Ranger, and CellBender are installed and on your PATH.

export COMBINED_OUTPUT_TYPE=anndata  # or 'sce'
snakemake --cores 8

---

📊 Outputs

• QC plots: results/r_analysis/qc_report.pdf
• Clustered Seurat object: combined_seurat.rds
• Markers table: markers.csv
• Optional: combined.rds (SCE) or combined.h5ad (AnnData)

---

🧪 Pipeline Steps

1. FastQC (raw)
2. Cell Ranger alignment
3. CellBender background removal
4. FastQC (post-cleaning)
5. scDblFinder doublet prediction
6. Seurat QC + UMAP + clustering
7. Merge & output: AnnData / SCE
8. Clustering & marker detection

---

🔁 Customization

To change final output format:

export COMBINED_OUTPUT_TYPE=sce    # or "anndata"

To change clustering resolution or genes used for PCA, edit scRNA_qc_plots.R.

---

📫 Contact

Tawaun Lucas
Postdoctoral Fellow, Genentech
tawanl@gmail.com