Single-Cell Variant Calling Pipeline

This repository contains a modular pipeline for variant calling on single-cell RNA-seq data, using Cell Ranger, a Python-based BAM splitting tool, and cellsnp-lite.

Features

Compatible with 10x Genomics (Cell Ranger) or Drop-seq BAMs.
Automatically splits pooled BAM into per-cell BAMs.
Efficient SNP/variant calling per cell using cellsnp-lite.
User-friendly: prompts for inputs, no hardcoded personal paths.

Requirements

Cell Ranger
Python 3.6+ with pysam
samtools
cellsnp-lite
Unix/Linux shell

Reference File Preparation

You will need:

Reference genome directory (e.g., from 10x Genomics or built via cellranger mkref)
SNP VCF file (e.g., dbSNP or 1000 Genomes, bgzipped and tabix-indexed)

Example commands (for human):

# Download Cell Ranger reference (GRCh38)
wget https://cf.10xgenomics.com/supp/cell-exp/refdata-gex-GRCh38-2020-A.tar.gz
tar -xzf refdata-gex-GRCh38-2020-A.tar.gz

# Download and index SNP VCF
wget https://ftp.ncbi.nih.gov/snp/latest_release/VCF/GCF_000001405.39.gz
tabix -p vcf GCF_000001405.39.gz

For other species, obtain reference and SNP VCFs from Ensembl, NCBI, or appropriate consortia.
Set the reference genome path in Cell Ranger (--transcriptome=...).
Provide the SNP VCF path to the pipeline when prompted.

Processing Barcoded Single-Cell Data (10x, Drop-seq)

For data containing single-cell barcodes (e.g., 10x Genomics, Drop-seq), use our integrated SLURM pipeline:

Introduction: scRNA_VariantCalling_PipelineV2.0.sh

scRNA_VariantCalling_PipelineV2.0.sh is an end-to-end pipeline that takes raw FASTQ files through alignment (using Cell Ranger or STARsolo), performs quality control, and runs per-cell variant calling using cellsnp-lite.

It's designed for use on HPC clusters with SLURM and supports 10x Genomics, Drop-seq, and plate-based protocols.

Key Features

detects protocol: Supports 10x Genomics, Drop-Seq, and SMART-Seq2/plate data.
End-to-end: Handles FastQC, alignment, barcode handling, and per-cell variant calling.
No hardcoded paths or user info: Adaptable for any lab or HPC setup.
Compatible with modular environments and conda.

Usage

Submit your job (edit the script to match your paths and reference files):

sbatch scRNA_VariantCalling_PipelineV2.0.sh <SAMPLE_ID> <PROTOCOL>

Processing sc-RNA without barcode

1. Process FASTQ with Cell Ranger or any alignment tools

Generate a coordinate-sorted BAM file with Tag:

cellranger count \
  --id=YOUR_SAMPLE_ID \
  --transcriptome=/path/to/reference \
  --fastqs=/path/to/fastqs \
  --sample=YOUR_SAMPLE_NAME

2. Run the Split BAM by Cell Barcode & Variant Calling Pipeline

Run the bash pipeline:

bash bam2snp.sh

The first part of the pipeline contains the script that split bam by cell barcode

python splitbam.py

The script will prompt for:

Sample name (for output labeling)
Path to input BAM (e.g., Cell Ranger’s possorted_genome_bam.bam)
Barcode tag (CB for 10x, XC for Drop-seq)
Minimum reads per barcode to keep

The second half picks up the output of bamsplit.py and perform variant calling with cellsnp-lite

The script will prompt for:

Output directories for per-cell BAMs and SNP results
Reference SNP VCF (e.g., dbSNP)

4. Results

Variant calls per cell will be in the specified output directory and ready for downstream analysis.

Notes

Indexing of per-cell BAMs is automatic.
Intermediate files (e.g., list files) are cleaned up at the end of the run.
To save storage, you can optionally delete per-cell BAMs after variant calling.

Citing Tools

Please cite:

License

See LICENSE for terms.

Contact

For questions, open a GitHub issue or contact the repository maintainer.

YanshengLuo/singlecell-variantcalling-pipeline