MACS2

Overview

MACS2 (Model-based Analysis of ChIP-Seq) is the standard peak caller for ChIP-seq and ATAC-seq experiments. It identifies regions of the genome where sequencing reads are significantly enriched compared to background, modelling the shift between forward and reverse strand reads to improve spatial resolution. MACS2 supports both narrow peak calling (e.g., transcription factor binding sites, ATAC-seq open chromatin) and broad peak calling (e.g., histone modifications like H3K27me3 or H3K36me3). It uses a dynamic Poisson distribution to capture local biases in the genome and reports peaks with associated p-values and q-values.

Installation

mamba install -c bioconda macs2

Basic Usage

ATAC-seq peak calling

For ATAC-seq data, use --nomodel with --shift and --extsize to centre reads on the Tn5 cut site.

macs2 callpeak -t sample.filtered.bam \
  -f BAMPE -g hs \
  --nomodel --shift -100 --extsize 200 \
  -n sample --outdir macs2_output/ \
  -q 0.01

ChIP-seq peak calling with input control

macs2 callpeak -t treatment.bam -c input.bam \
  -f BAMPE -g hs \
  -n chip_sample --outdir macs2_output/ \
  -q 0.05

Broad peak calling (e.g., H3K27me3)

macs2 callpeak -t treatment.bam -c input.bam \
  -f BAMPE -g hs --broad \
  -n broad_sample --outdir macs2_output/

Key Parameters

Flag / option	Description
`-t`	Treatment (enriched) BAM file.
`-c`	Control (input) BAM file. Optional for ATAC-seq; recommended for ChIP-seq.
`-f`	Input format (`BAMPE` for paired-end BAM, `BAM` for single-end). `BAMPE` uses actual fragment sizes from paired reads.
`-g`	Effective genome size (`hs` for human, `mm` for mouse, or a numeric value).
`-n`	Name prefix for output files.
`--outdir`	Output directory for results.
`-q`	Minimum FDR (q-value) cutoff for peak detection (default: 0.05).
`--nomodel`	Do not build the shifting model; use `--extsize` instead. Required for ATAC-seq.
`--shift`	Shift read positions by this many bases (use `-100` with `--extsize 200` for ATAC-seq to centre on cut sites).
`--extsize`	Extend reads to this fragment size when `--nomodel` is set.
`--broad`	Call broad peaks (gapped peaks) instead of narrow peaks. Use for diffuse histone marks.

Expected Output

For narrow peak calling:

sample_peaks.narrowPeak – BED6+4 format file listing each peak with chromosome, start, end, name, score, strand, signal value, p-value, q-value, and summit offset.
sample_summits.bed – BED file with the single-base-pair summit position of each peak.
sample_peaks.xls – tab-delimited spreadsheet with detailed peak statistics.
sample_treat_pileup.bdg – treatment pileup signal in bedGraph format.
sample_control_lambda.bdg – local background lambda in bedGraph format.

For broad peak calling, sample_peaks.broadPeak and sample_peaks.gappedPeak replace the narrowPeak output.