scATAC-Seq Analysis Pipeline

This repository provides a Snakemake pipeline for analyzing single-cell ATAC-seq (scATAC-seq) data using various tools, including cellranger-atac, Socrates, Genrich, and bedtools. The pipeline processes raw scATAC fastq files, performs barcode cleaning, demultiplexes cells, and identifies accessible chromatin regions (ACRs).

Table of Contents

• Installation
• Configuration
• Usage
• Pipeline Steps
• Outputs

Installation

Prerequisites

• Install Snakemake
• Required software tools: cellranger-atac, sinto, samtools, R, popscle, bedtools, and ncbi-blast
• Ensure access to SLURM for resource allocation on a high-performance computing (HPC) cluster.

Set up Conda environments

For dependencies, create Conda environments as listed in config/SocratesEnv.yaml and config/DemuxletEnv.yaml conda env create -f config/SocratesEnv.yaml conda env create -f config/DemuxletEnv.yaml

Configuration

Edit paths in config/config.yaml

for cellranger

1. CELLRANGER_PATH: path to cellranger-atac software
2. CELLRANGER_ref: path to reference created with cellranger
3. scATACraw: path to fastq of single-cell ATAC-seq reads
4. NAME: Sample name of reference for cellranger-atac
5. Nuclear: expression for scaffold prefix for nuclear genome for sinto usage
6. Plastid: expression for scaffold prefix for plastid genome
7. GFF: gene annotation in gff format
8. CHRFILE: lengths of scaffolds

for demultiplexing

9. MACSpath: path to macs2 software
10. VCFdemux: path to VCF for demuxlet
11. SAMPLENAMES: file with atac-seq sample names
12. PICARD: path to picard software

for ACR calling

13. GENRICH: path to genrich software
14. SAMPLEMETA: "Metadata/SampleNames.txt"
15. WGS: "/path/to/WGS_alignments"
16. REFERENCE: "/path/to/reference_genome.fa"
17. PLASTID: "/path/to/plastid_blast_db"

Pipeline Steps

1. Cell Ranger ATAC Processing Description: Runs cellranger-atac to process raw scATAC fastq files and generate BAM and fragment files. Input: Raw fastq files, reference genome. Output: {NAME}/outs/possorted_bam.bam, {NAME}/outs/fragments.tsv
2. Cell Barcode Cleaning Description: Formats and filters cell barcodes using the sinto tool. Input: BAM and fragment files from Cell Ranger ATAC output. Output: Filtered fragments and contig lists.
3. Cell Identification (Socrates) Description: Uses Socrates to filter cell barcodes based on genomic regions. Output: Filtered barcodes and sparse matrices for further analysis.
4. Demultiplexing Cells Description: Uses demuxlet with SNP data to demultiplex cells. Input: BAM file, VCF file of SNPs, sample names. Output: Best match demultiplex results, clean barcode list.
5. Filtering Barcodes Description: Filters barcodes using picard to remove duplicates and produce a final BAM. Output: Filtered and indexed BAM files.
6. Downsample Nuclei Description: Randomly downsamples barcodes to a specified count (e.g., 552 per sample). Input: List of clean barcodes. Output: Downsampled barcode files.
7. ACR Calling Description: Uses Genrich to call Accessible Chromatin Regions (ACRs) based on downsampled BAM files. Output: Peak files and ACR log files.
8. ACR Processing Description: Filters out regions aligning to plastid genomes and calculates jaccard similarity. Output: Processed and concatenated ACR files for frequency analysis.
9. ACR Analysis and Visualization Description: Analyzes and visualizes ACR positions relative to genes. Output: Merged and classified ACR files, visualization plots.
10. ACR Classification Description: Classifies ACRs based on proximity to genes and exons. Output: ACRs classified by gene location, overlap with exons, and nearest gene.

Outputs

The pipeline produces a variety of output files, including:

Filtered BAM files: Located in DEMUX/BAMscATAC/ Downsampled ACR files: Stored in DownsampledBams/Peaks/ Merged ACRs: Final classified ACR files for downstream analysis Visualization files: Plots summarizing ACRs ACRs classified by genomic context ACRs_classified/