本文へ移動
cccskills
無料GitHub で公開

bio-chipseq-peak-calling

ChIP-seq peak calling using MACS3 (or MACS2). Call narrow peaks for transcription factors or broad peaks for histone modifications. Supports input control, fragment size modeling, and various output formats including narrowPeak and broadPeak BED files. Use when calling peaks from ChIP-seq alignments.

インストール方法を見る

含まれるファイル(3)

  • SKILL.md8.0 KB
  • examples/macs3_peak_calling.sh998 B
  • usage-guide.md1.7 KB

SKILL.md(原文)

インストールする前に、エージェントに与えられる指示の中身を確認できます。

Version Compatibility

Reference examples tested with: MACS2 2.2+, MACS3 3.0+

Before using code patterns, verify installed versions match. If versions differ:

  • CLI: <tool> --version then <tool> --help to confirm flags

If code throws ImportError, AttributeError, or TypeError, introspect the installed package and adapt the example to match the actual API rather than retrying.

Peak Calling with MACS3

"Call peaks from my ChIP-seq data" → Identify significantly enriched regions (narrow peaks for TFs, broad peaks for histone marks) by comparing IP signal to input control.

  • CLI: macs3 callpeak -t chip.bam -c input.bam -f BAM -g hs -n sample

MACS3 is the actively developed successor to MACS2. Commands are identical except the binary name. MACS2 is in maintenance mode.

Basic Peak Calling

Goal: Call enriched regions from ChIP-seq alignments with input control normalization.

Approach: Compare treatment BAM signal against input control using MACS3 local Poisson model.

# Call peaks with input control (recommended)
macs3 callpeak -t chip.bam -c input.bam -f BAM -g hs -n sample --outdir peaks/

# For MACS2 (legacy), replace 'macs3' with 'macs2' - syntax is identical

Without Input Control

Goal: Call peaks without a matched input/control sample.

Approach: Use MACS3 with genomic background estimation only (less accurate than with control).

# Not recommended, but possible
macs3 callpeak -t chip.bam -f BAM -g hs -n sample --outdir peaks/

Narrow Peaks (TF, H3K4me3, H3K27ac)

Goal: Call sharp, well-defined peaks typical of transcription factors and active histone marks.

Approach: Use default narrow peak mode with q-value filtering and genome size correction.

macs3 callpeak \
    -t chip.bam \
    -c input.bam \
    -f BAM \
    -g hs \                        # hs=human, mm=mouse, ce=worm, dm=fly
    -n sample_narrow \
    --outdir peaks/ \
    -q 0.05                        # q-value threshold

Broad Peaks (H3K36me3, H3K27me3, H3K9me3)

Goal: Call diffuse, broad enrichment domains typical of repressive or elongation-associated histone marks.

Approach: Enable broad peak mode which links nearby enriched regions into broader domains.

macs3 callpeak \
    -t chip.bam \
    -c input.bam \
    -f BAM \
    -g hs \
    -n sample_broad \
    --outdir peaks/ \
    --broad \                      # Broad peak mode
    --broad-cutoff 0.1             # Broad peak q-value

Paired-End Data

Goal: Call peaks from paired-end sequencing using actual fragment sizes instead of modeled estimates.

Approach: Use BAMPE format so MACS3 calculates fragment size from mate pairs directly.

# MACS3 uses BAMPE format for paired-end
macs3 callpeak \
    -t chip.bam \
    -c input.bam \
    -f BAMPE \                     # Paired-end BAM
    -g hs \
    -n sample_pe \
    --outdir peaks/

Multiple Replicates

Goal: Call peaks from multiple biological replicates pooled together for increased statistical power.

Approach: Provide all replicate BAMs to MACS3, which internally pools reads before peak calling.

# Pool replicates (MACS3 handles internally)
macs3 callpeak \
    -t rep1.bam rep2.bam rep3.bam \
    -c input.bam \
    -f BAM \
    -g hs \
    -n pooled \
    --outdir peaks/

Custom Genome Size

Goal: Call peaks for non-model organisms without a built-in genome size shortcut.

Approach: Provide the effective genome size as a numeric value instead of a species abbreviation.

# For non-model organisms or custom genomes
macs3 callpeak \
    -t chip.bam \
    -c input.bam \
    -f BAM \
    -g 2.7e9 \                     # Effective genome size in bp
    -n sample \
    --outdir peaks/

Common Genome Sizes

GenomeFlagEffective Size
Humanhs2.7e9
Mousemm1.87e9
C. elegansce9e7
D. melanogasterdm1.2e8

Fixed Fragment Size

Goal: Call peaks when fragment size modeling fails or a specific extension size is needed.

Approach: Bypass model building and specify a fixed read extension size manually.

# If modeling fails or for ATAC-seq
macs3 callpeak \
    -t chip.bam \
    -c input.bam \
    -f BAM \
    -g hs \
    --nomodel \                    # Skip model building
    --extsize 200 \                # Fixed extension size
    -n sample \
    --outdir peaks/

Generate Signal Tracks

Goal: Produce normalized signal tracks for genome browser visualization alongside peak calls.

Approach: Enable bedGraph output with signal-per-million-reads normalization, then convert to bigWig.

# Generate bedGraph and bigWig files
macs3 callpeak \
    -t chip.bam \
    -c input.bam \
    -f BAM \
    -g hs \
    -n sample \
    --outdir peaks/ \
    -B \                           # Generate bedGraph
    --SPMR                         # Signal per million reads

# Convert to bigWig (requires UCSC tools)
sort -k1,1 -k2,2n peaks/sample_treat_pileup.bdg > peaks/sample.sorted.bdg
bedGraphToBigWig peaks/sample.sorted.bdg chrom.sizes peaks/sample.bw

Local Lambda for Broad Marks

Goal: Improve broad peak calling by disabling the genome-wide lambda estimate.

Approach: Use --nolambda to rely solely on local background estimation for very broad domains.

# Recommended for very broad marks
macs3 callpeak \
    -t chip.bam \
    -c input.bam \
    -f BAM \
    -g hs \
    --broad \
    --nolambda \                   # Use local lambda only
    -n sample \
    --outdir peaks/

Cutoff Analysis

Goal: Evaluate how different significance thresholds affect the number of called peaks.

Approach: Run MACS3 cutoff analysis mode to generate a table of peak counts at various q-value cutoffs.

# Test different q-value cutoffs
macs3 callpeak \
    -t chip.bam \
    -c input.bam \
    -f BAM \
    -g hs \
    --cutoff-analysis \            # Generate cutoff analysis file
    -n sample \
    --outdir peaks/

Output Files

FileDescription
*_peaks.narrowPeakPeak coordinates (BED6+4)
*_peaks.broadPeakBroad peak coordinates
*_summits.bedPeak summit positions
*_model.rR script for model visualization
*_treat_pileup.bdgTreatment signal (with -B)
*_control_lambda.bdgControl signal (with -B)

narrowPeak Format

chr1  100  200  peak_1  100  .  5.2  10.5  8.3  50

Columns: chr, start, end, name, score, strand, signalValue, pValue, qValue, peak

Filter Peaks

Goal: Post-filter called peaks by statistical significance or signal strength.

Approach: Use awk on narrowPeak columns to apply q-value or signal-value cutoffs.

# Filter by q-value
awk '$9 > 2' peaks.narrowPeak > peaks.filtered.narrowPeak  # -log10(q) > 2 means q < 0.01

# Sort by signal strength
sort -k7,7nr peaks.narrowPeak > peaks.sorted.narrowPeak

Key Parameters

ParameterDefaultDescription
-trequiredTreatment BAM file(s)
-cnoneControl BAM file(s)
-fAUTOFormat (BAM, BAMPE, BED)
-ghsGenome size
-nNAOutput prefix
-q0.05Q-value cutoff
-pnoneP-value cutoff (overrides -q)
--broadfalseBroad peak calling
--nomodelfalseSkip model building
--extsize200Extension size (with --nomodel)
-BfalseGenerate bedGraph
--SPMRfalseSignal per million reads

Related Skills

  • peak-annotation - Annotate peaks to genes
  • differential-binding - Compare peaks between conditions
  • alignment-files - Prepare BAM files
  • chipseq-visualization - Visualize peaks

レビュー

まだレビューはありません。使ってみた感想をお寄せください。

同じリポジトリのスキル

概要と使いどころ

adaptyv

無料

Cloud laboratory platform for automated protein testing and validation. Use when designing proteins and needing experimental validation including binding assays, expression testing, thermostability measurements, enzyme activity assays, or protein sequence optimization. Also use for submitting experiments via API, tracking experiment status, downloading results, optimizing protein sequences for better expression using computational tools (NetSolP, SoluProt, SolubleMPNN, ESM), or managing protein design workflows with wet-lab validation.

日本語の概要は準備中です。原文の説明を表示しています。

FreedomIntelligence/OpenClaw-Medical-Skills3,0622026年7月21日 更新

Time-blind friendly planning, executive function support, and daily structure for ADHD brains. Specializes in realistic time estimation, dopamine-aware task design, and building systems that actually work for neurodivergent minds.

日本語の概要は準備中です。原文の説明を表示しています。

FreedomIntelligence/OpenClaw-Medical-Skills3,0622026年7月21日 更新

aeon

無料

This skill should be used for time series machine learning tasks including classification, regression, clustering, forecasting, anomaly detection, segmentation, and similarity search. Use when working with temporal data, sequential patterns, or time-indexed observations requiring specialized algorithms beyond standard ML approaches. Particularly suited for univariate and multivariate time series analysis with scikit-learn compatible APIs.

日本語の概要は準備中です。原文の説明を表示しています。

FreedomIntelligence/OpenClaw-Medical-Skills3,0622026年7月21日 更新

Browse the web for any task — research topics, read articles, interact with web apps, fill forms, take screenshots, extract data, and test web pages. Use whenever a browser would be useful, not just when the user explicitly asks.

日本語の概要は準備中です。原文の説明を表示しています。

FreedomIntelligence/OpenClaw-Medical-Skills3,0622026年7月21日 更新

AI驱动的综合健康分析系统,整合多维度健康数据、识别异常模式、预测健康风险、提供个性化建议。支持智能问答和AI健康报告生成。

日本語の概要は準備中です。原文の説明を表示しています。

FreedomIntelligence/OpenClaw-Medical-Skills3,0622026年7月21日 更新

alphafold

無料

Validate protein designs using AlphaFold2 structure prediction. Use this skill when: (1) Validating designed sequences fold correctly, (2) Predicting binder-target complex structures, (3) Calculating confidence metrics (pLDDT, pTM, ipTM), (4) Self-consistency validation of designs, (5) Multi-chain complex prediction with AlphaFold-Multimer. For faster single-chain prediction, use esm. For QC thresholds, use protein-qc.

日本語の概要は準備中です。原文の説明を表示しています。

FreedomIntelligence/OpenClaw-Medical-Skills3,0622026年7月21日 更新

FreedomIntelligence のスキルをすべて見る

このスキルの問題を報告する