Skip to content

Latest commit

 

History

History
90 lines (64 loc) · 2.57 KB

File metadata and controls

90 lines (64 loc) · 2.57 KB

ATAC-seq Pipeline - Usage Guide

Overview

This workflow processes ATAC-seq data from raw FASTQ files to accessibility peaks, with optional differential analysis and transcription factor footprinting.

Prerequisites

# CLI tools
conda install -c bioconda fastp bowtie2 samtools macs3 deeptools bedtools tobias

# R packages
BiocManager::install(c('DiffBind', 'ChIPseeker'))

Quick Start

Tell your AI agent what you want to do:

  • "Run the ATAC-seq pipeline on my samples"
  • "Call accessibility peaks from my ATAC-seq data"
  • "Find differential accessibility between treatment and control"

Example Prompts

Starting from FASTQ

"Process my ATAC-seq FASTQ files through peak calling"

"Run ATAC-seq analysis on human samples"

"I have paired-end ATAC-seq, align and call peaks"

Analysis

"Calculate TSS enrichment for my ATAC-seq"

"Find differential peaks between conditions"

"Run TF footprinting with TOBIAS"

Input Requirements

Input Format Description
FASTQ files .fastq.gz Paired-end reads
Reference FASTA Reference genome + Bowtie2 index
Motifs (optional) JASPAR For footprinting analysis

What the Workflow Does

  1. Quality Control - Trim Nextera adapters
  2. Alignment - Map reads with Bowtie2
  3. BAM Processing - Remove chrM, shift for Tn5, deduplicate
  4. Peak Calling - Call accessible regions with MACS3
  5. QC - TSS enrichment, FRiP, fragment sizes
  6. Differential - Compare accessibility between conditions
  7. Footprinting - Infer TF binding from accessibility patterns

ATAC-seq vs ChIP-seq Processing

Aspect ATAC-seq ChIP-seq
Adapters Nextera TruSeq
Control None needed Input required
Tn5 shift Yes (+4/-5 bp) No
chrM High, remove Low
Peak type Narrow Narrow or broad

Tips

  • Mitochondrial: Expect 20-50% chrM reads; always filter
  • Tn5 shift: Essential for accurate footprinting
  • TSS enrichment: Good library shows >5 enrichment
  • Fragment sizes: Should show nucleosome-free and nucleosome peaks
  • Footprinting: Requires high depth (>50M reads)