MACS: Model-based Analysis for ChIP-Seq

Latest Release:

• Github:
• PyPI:
• Bioconda:
• Debian Med:

Introduction

With the improvement of sequencing techniques, chromatin immunoprecipitation followed by high throughput sequencing (ChIP-Seq) is getting popular to study genome-wide protein-DNA interactions. To address the lack of powerful ChIP-Seq analysis method, we presented the Model-based Analysis of ChIP-Seq (MACS), for identifying transcript factor binding sites. MACS captures the influence of genome complexity to evaluate the significance of enriched ChIP regions and MACS improves the spatial resolution of binding sites through combining the information of both sequencing tag position and orientation. MACS can be easily used for ChIP-Seq data alone, or with a control sample with the increase of specificity. Moreover, as a general peak-caller, MACS can also be applied to any "DNA enrichment assays" if the question to be asked is simply: where we can find significant reads coverage than the random background.

Note: new development on MACS has been moved to MACS3 repository, we will only fix installation issues on MACS2.

Recent Changes for MACS (2.2.9.1)

2.2.9.1

* Bug fix:
 
Cython has a major upgrade to 3.0, and can't work directly with
MACS2 codes, so we changed the requirement for Cython to 0.29.*.

Python support has been changed to 3.7 to 3.11. Numpy >=0.19.

We tested MACS2 on Mac OS12 through Github Actions as well.

Thank @jemajet! #569

2.2.8

* Bug fix:

MACS2 typo in 'setup.py': 'numpy>=>=1.17' -> 'numpy>=1.17'
PR #543, issues #535, #541, #544

Now test on and support Python 3.6/3.7/3.8/3.9/3.10/3.11

2.2.7.1

* hotfix: 

Add 'wheel' and 'pip' to pyproject.toml so that `pip install` can 
work.

2.2.7

* Bugs fixed

1) MACS2 has been tested on multiple architectures to make sure it
can successfully generate consistent results. Currently the
supported architectures are: AMD64, ARM64, i386, PPC64LE, and
S390X. Thanks to @mr-c, @junaruga, and @tillea! Related to issue
#340, #349, #351, and #359; to PR #348, #350, #360, #361, #367,
and #370. The lesson is that if the project is built on Cython and
is aimed at memory efficiency, we should specifically define all
int/float types in pyx files such as int8_t or uint32_t using
either libc or numpy (c version) instead of relying on Cython
types such as short, long, double.

2) MACS2 setup script will check numpy and install numpy if
necessary. PR #378, issue #364

3) `bdgbroadcall` command will correctly add the score column (5th
column). The score (5th) column contains 10 times of the average
score in the broad region. PR #373, issue #362

4) The missing test on `bdgopt` subcommand has been added. PR #363

5) The obsolete option `--ratio` from `callpeak` subcommand has
been removed. PR #369, issue #366

6) Fixed the incorrect description in README on the 'maximum
length of broad region is 4 times of d' to 'maximum gap for
merging broad regions is 4 times of tag size by default'. PR #380,
issue #365.

* Other

1) CODE OF CONDUCT document has been added to MACS2 github
repository. PR #358

2.2.6

* New Features 

1) Speed up MACS2. Some programming tricks and code cleanup. The 
filter_dup function replaces separate_dups. The later one was 
implemented for potentially putting back duplicate reads in 
certain downstream analysis. However such analysis hasn't been 
implemented. Optimize speed of writing bedGraph files. Optimize 
BAM and BAMPE parsing with pointer casting instead of python 
unpack. 

2) The comment lines in the headers of BED or SAM files will be
correctly skipped. However, MACS2 won't check comment lines in the
middle of the file.

* Bugs fixed 

1) Cutoff-analysis in callpeak command. #341

2) Issues related to SAMParser and three ELAND Parsers are
fixed. #347

* Other 

1) cmdlinetest script in test/ folder has been updated to: 1. test 
cutoff-analysis with callpeak cmd; 2. output the 2 lines before 
and after the error or warning message during tests; 3. output 
only the first 10 lines if the difference between test result and 
standard result can be found; 4. prockreport monitor CPU time and 
memory usage in 1 sec interval -- a bit more accurate.

2) Python3.5 support is removed. Now MACS2 requires Python>=3.6.

Install

Please check the file 'INSTALL.md' in the distribution. In general, you can install through PyPI as pip install macs2. To use virtual environment is highly recommended. Or you can install after unzipping the released package downloaded from Github, then use pip install . command.

Usage

macs2 [-h] [--version]
    {callpeak,bdgpeakcall,bdgbroadcall,bdgcmp,bdgopt,cmbreps,bdgdiff,filterdup,predictd,pileup,randsample,refinepeak}

Example for regular peak calling: macs2 callpeak -t ChIP.bam -c Control.bam -f BAM -g hs -n test -B -q 0.01

Example for broad peak calling: macs2 callpeak -t ChIP.bam -c Control.bam --broad -g hs --broad-cutoff 0.1

There are twelve functions available in MACS2 serving as sub-commands.

Subcommand	Description
callpeak	Main MACS2 Function to call peaks from alignment results.
bdgpeakcall	Call peaks from bedGraph output.
bdgbroadcall	Call broad peaks from bedGraph output.
bdgcmp	Comparing two signal tracks in bedGraph format.
bdgopt	Operate the score column of bedGraph file.
cmbreps	Combine BEDGraphs of scores from replicates.
bdgdiff	Differential peak detection based on paired four bedGraph files.
filterdup	Remove duplicate reads, then save in BED/BEDPE format.
predictd	Predict d or fragment size from alignment results.
pileup	Pileup aligned reads (single-end) or fragments (paired-end)
randsample	Randomly choose a number/percentage of total reads.
refinepeak	Take raw reads alignment, refine peak summits.

We only cover callpeak subcommand in this document. Please use macs2 COMMAND -h to see the detail description for each option of each subcommand.

Call peaks

This is the main function in MACS2. It can be invoked by macs2 callpeak . If you type this command with -h, you will see a full description of command-line options. Here we only list the essentials.

Essential Options

-t/--treatment FILENAME

This is the only REQUIRED parameter for MACS. The file can be in any supported format -- see detail in the --format option. If you have more than one alignment file, you can specify them as -t A B C. MACS will pool up all these files together.

-c/--control

The control, genomic input or mock IP data file. Please follow the same direction as for -t/--treatment.

-n/--name

The name string of the experiment. MACS will use this string NAME to create output files like NAME_peaks.xls, NAME_negative_peaks.xls, NAME_peaks.bed , NAME_summits.bed, NAME_model.r and so on. So please avoid any confliction between these filenames and your existing files.

--outdir

MACS2 will save all output files into the specified folder for this option. A new folder will be created if necessary.

-f/--format FORMAT

Format of tag file can be ELAND, BED, ELANDMULTI, ELANDEXPORT, SAM, BAM, BOWTIE, BAMPE, or BEDPE. Default is AUTO which will allow MACS to decide the format automatically. AUTO is also useful when you combine different formats of files. Note that MACS can't detect BAMPE or BEDPE format with AUTO, and you have to implicitly specify the format for BAMPE and BEDPE.

Nowadays, the most common formats are BED or BAM (including BEDPE and BAMPE). Our recommendation is to convert your data to BED or BAM first.

Also, MACS2 can detect and read gzipped file. For example, .bed.gz file can be directly used without being uncompressed with --format BED.

Here are detailed explanation of the recommanded formats:

BED

The BED format can be found at UCSC genome browser website.

The essential columns in BED format input are the 1st column chromosome name, the 2nd start position, the 3rd end position, and the 6th, strand.

Note that, for BED format, the 6th column of strand information is required by MACS. And please pay attention that the coordinates in BED format are zero-based and half-open. See more detail at UCSC site.

BAM/SAM

If the format is BAM/SAM, please check the definition in (http://samtools.sourceforge.net/samtools.shtml). If the BAM file is generated for paired-end data, MACS will only keep the left mate(5' end) tag. However, when format BAMPE is specified, MACS will use the real fragments inferred from alignment results for reads pileup.

BEDPE or BAMPE

A special mode will be triggered while the format is specified as BAMPE or BEDPE. In this way, MACS2 will process the BAM or BED files as paired-end data. Instead of building a bimodal distribution of plus and minus strand reads to predict fragment size, MACS2 will use actual insert sizes of pairs of reads to build fragment pileup.

The BAMPE format is just a BAM format containing paired-end alignment information, such as those from BWA or BOWTIE.

The BEDPE format is a simplified and more flexible BED format, which only contains the first three columns defining the chromosome name, left and right position of the fragment from Paired-end sequencing. Please note, this is NOT the same format used by BEDTOOLS, and the BEDTOOLS version of BEDPE is actually not in a standard BED format. You can use MACS2 subcommand randsample to convert a BAM file containing paired-end information to a BEDPE format file:

macs2 randsample -i the_BAMPE_file.bam -f BAMPE -p 100 -o the_BEDPE_file.bed

-g/--gsize

PLEASE assign this parameter to fit your needs!

It's the mappable genome size or effective genome size which is defined as the genome size which can be sequenced. Because of the repetitive features on the chromosomes, the actual mappable genome size will be smaller than the original size, about 90% or 70% of the genome size. The default hs -- 2.7e9 is recommended for human genome. Here are all precompiled parameters for effective genome size:

• hs: 2.7e9
• mm: 1.87e9
• ce: 9e7
• dm: 1.2e8

Users may want to use k-mer tools to simulate mapping of Xbps long reads to target genome, and to find the ideal effective genome size. However, usually by taking away the simple repeats and Ns from the total genome, one can get an approximate number of effective genome size. A slight difference in the number won't cause a big difference of peak calls, because this number is used to estimate a genome-wide noise level which is usually the least significant one compared with the local biases modeled by MACS.

-s/--tsize

The size of sequencing tags. If you don't specify it, MACS will try to use the first 10 sequences from your input treatment file to determine the tag size. Specifying it will override the automatically determined tag size.

-q/--qvalue

The q-value (minimum FDR) cutoff to call significant regions. Default is 0.05. For broad marks, you can try 0.05 as the cutoff. Q-values are calculated from p-values using the Benjamini-Hochberg procedure.

-p/--pvalue

The p-value cutoff. If -p is specified, MACS2 will use p-value instead of q-value.

--min-length, --max-gap

These two options can be used to fine-tune the peak calling behavior by specifying the minimum length of a called peak and the maximum allowed a gap between two nearby regions to be merged. In other words, a called peak has to be longer than min-length, and if the distance between two nearby peaks is smaller than max-gap then they will be merged as one. If they are not set, MACS2 will set the DEFAULT value for min-length as the predicted fragment size d, and the DEFAULT value for max-gap as the detected read length. Note, if you set a min-length value smaller than the fragment size, it may have NO effect on the result. For broad peak calling with --broad option set, the DEFAULT max-gap for merging nearby stronger peaks will be the same as narrow peak calling, and 4 times of the max-gap will be used to merge nearby weaker (broad) peaks. You can also use --cutoff-analysis option with the default setting, and check the column avelpeak under different cutoff values to decide a reasonable min-length value.

--nolambda

With this flag on, MACS will use the background lambda as local lambda. This means MACS will not consider the local bias at peak candidate regions.

--slocal, --llocal

These two parameters control which two levels of regions will be checked around the peak regions to calculate the maximum lambda as local lambda. By default, MACS considers 1000bp for small local region(--slocal), and 10000bps for large local region(--llocal) which captures the bias from a long-range effect like an open chromatin domain. You can tweak these according to your project. Remember that if the region is set too small, a sharp spike in the input data may kill a significant peak.

--nomodel

While on, MACS will bypass building the shifting model.

--extsize

While --nomodel is set, MACS uses this parameter to extend reads in 5'->3' direction to fix-sized fragments. For example, if the size of the binding region for your transcription factor is 200 bp, and you want to bypass the model building by MACS, this parameter can be set as 200. This option is only valid when --nomodel is set or when MACS fails to build model and --fix-bimodal is on.

--shift

Note, this is NOT the legacy --shiftsize option which is replaced by --extsize! You can set an arbitrary shift in bp here. Please Use discretion while setting it other than the default value (0). When --nomodel is set, MACS will use this value to move cutting ends (5') then apply --extsize from 5' to 3' direction to extend them to fragments. When this value is negative, ends will be moved toward 3'->5' direction, otherwise 5'->3' direction. Recommended to keep it as default 0 for ChIP-Seq datasets, or -1 * half of EXTSIZE together with --extsize option for detecting enriched cutting loci such as certain DNAseI-Seq datasets. Note, you can't set values other than 0 if the format is BAMPE or BEDPE for paired-end data. The default is 0.

Here are some examples for combining --shift and --extsize:

1. To find enriched cutting sites such as some DNAse-Seq datasets. In this case, all 5' ends of sequenced reads should be extended in both directions to smooth the pileup signals. If the wanted smoothing window is 200bps, then use --nomodel --shift -100 --extsize 200.

2. For certain nucleosome-seq data, we need to pile up the centers of nucleosomes using a half-nucleosome size for wavelet analysis (e.g. NPS algorithm). Since the DNA wrapped on nucleosome is about 147bps, this option can be used: --nomodel --shift 37 --extsize 73.

--keep-dup

It controls the MACS behavior towards duplicate tags at the exact same location -- the same coordination and the same strand. The default auto option makes MACS calculate the maximum tags at the exact same location based on binomial distribution using 1e-5 as p-value cutoff; and the all option keeps every tag. If an integer is given, at most this number of tags will be kept at the same location. The default is to keep one tag at the same location. Default: 1

--broad

When this flag is on, MACS will try to composite broad regions in BED12 ( a gene-model-like format ) by putting nearby highly enriched regions into a broad region with loose cutoff. The broad region is controlled by another cutoff through --broad-cutoff. Please note that, the max-gap value for merging nearby weaker/broad peaks is 4 times of max-gap for merging nearby stronger peaks. The later one can be controlled by --max-gap option, and by default it is the average fragment/insertion length in the PE data. DEFAULT: False

--broad-cutoff

Cutoff for the broad region. This option is not available unless --broad is set. If -p is set, this is a p-value cutoff, otherwise, it's a q-value cutoff. DEFAULT: 0.1

--scale-to

When set to large, linearly scale the smaller dataset to the same depth as the larger dataset. By default or being set as small, the larger dataset will be scaled towards the smaller dataset. Beware, to scale up small data would cause more false positives.

-B/--bdg

If this flag is on, MACS will store the fragment pileup, control lambda in bedGraph files. The bedGraph files will be stored in the current directory named NAME_treat_pileup.bdg for treatment data, NAME_control_lambda.bdg for local lambda values from control.

--call-summits

MACS will now reanalyze the shape of signal profile (p or q-score depending on the cutoff setting) to deconvolve subpeaks within each peak called from the general procedure. It's highly recommended to detect adjacent binding events. While used, the output subpeaks of a big peak region will have the same peak boundaries, and different scores and peak summit positions.

--buffer-size

MACS uses a buffer size for incrementally increasing internal array size to store reads alignment information for each chromosome or contig. To increase the buffer size, MACS can run faster but will waste more memory if certain chromosome/contig only has very few reads. In most cases, the default value 100000 works fine. However, if there are a large number of chromosomes/contigs in your alignment and reads per chromosome/contigs are few, it's recommended to specify a smaller buffer size in order to decrease memory usage (but it will take longer time to read alignment files). Minimum memory requested for reading an alignment file is about # of CHROMOSOME * BUFFER_SIZE * 8 Bytes. DEFAULT: 100000

Output files

1. NAME_peaks.xls is a tabular file which contains information about called peaks. You can open it in excel and sort/filter using excel functions. Information include:

   • chromosome name
   • start position of peak
   • end position of peak
   • length of peak region
   • absolute peak summit position
   • pileup height at peak summit
   • -log10(pvalue) for the peak summit (e.g. pvalue =1e-10, then this value should be 10)
   • fold enrichment for this peak summit against random Poisson distribution with local lambda,
   • -log10(qvalue) at peak summit

   Coordinates in XLS is 1-based which is different from BED format. When --broad is enabled for broad peak calling, the pileup, p-value, q-value, and fold change in the XLS file will be the mean value across the entire peak region, since peak summit won't be called in broad peak calling mode.

2. NAME_peaks.narrowPeak is BED6+4 format file which contains the peak locations together with peak summit, p-value, and q-value. You can load it to the UCSC genome browser. Definition of some specific columns are:

   • 5th: integer score for display. It's calculated as int(-10*log10pvalue) or int(-10*log10qvalue) depending on whether -p (pvalue) or -q (qvalue) is used as score cutoff. Please note that currently this value might be out of the [0-1000] range defined in UCSC ENCODE narrowPeak format. You can let the value saturated at 1000 (i.e. p/q-value = 10^-100) by using the following 1-liner awk: awk -v OFS="\t" '{$5=$5>1000?1000:$5} {print}' NAME_peaks.narrowPeak
   • 7th: fold-change at peak summit
   • 8th: -log10pvalue at peak summit
   • 9th: -log10qvalue at peak summit
   • 10th: relative summit position to peak start

   The file can be loaded directly to the UCSC genome browser. Remove the beginning track line if you want to analyze it by other tools.

3. NAME_summits.bed is in BED format, which contains the peak summits locations for every peak. The 5th column in this file is the same as what is in the narrowPeak file. If you want to find the motifs at the binding sites, this file is recommended. The file can be loaded directly to the UCSC genome browser. Remove the beginning track line if you want to analyze it by other tools.

4. NAME_peaks.broadPeak is in BED6+3 format which is similar to narrowPeak file, except for missing the 10th column for annotating peak summits. This file and the gappedPeak file will only be available when --broad is enabled. Since in the broad peak calling mode, the peak summit won't be called, the values in the 5th, and 7-9th columns are the mean value across all positions in the peak region. Refer to narrowPeak if you want to fix the value issue in the 5th column.

5. NAME_peaks.gappedPeak is in BED12+3 format which contains both the broad region and narrow peaks. The 5th column is the score for showing grey levels on the UCSC browser as in narrowPeak. The 7th is the start of the first narrow peak in the region, and the 8th column is the end. The 9th column should be RGB color key, however, we keep 0 here to use the default color, so change it if you want. The 10th column tells how many blocks including the starting 1bp and ending 1bp of broad regions. The 11th column shows the length of each block and 12th for the start of each block. 13th: fold-change, 14th: -log10pvalue, 15th: -log10qvalue. The file can be loaded directly to the UCSC genome browser. Refer to narrowPeak if you want to fix the value issue in the 5th column.

6. NAME_model.r is an R script which you can use to produce a PDF image of the model based on your data. Load it to R by:

   $ Rscript NAME_model.r

   Then a pdf file NAME_model.pdf will be generated in your current directory. Note, R is required to draw this figure.

7. The NAME_treat_pileup.bdg and NAME_control_lambda.bdg files are in bedGraph format which can be imported to the UCSC genome browser or be converted into even smaller bigWig files. The NAME_treat_pielup.bdg contains the pileup signals (normalized according to --scale-to option) from ChIP/treatment sample. The NAME_control_lambda.bdg contains local biases estimated for each genomic location from the control sample, or from treatment sample when the control sample is absent. The subcommand bdgcmp can be used to compare these two files and make a bedGraph file of scores such as p-value, q-value, log-likelihood, and log fold changes.

Other useful links

• Cistrome
• bedTools
• UCSC toolkits

Tips of fine-tuning peak calling

There are several subcommands within MACSv2 package to fine-tune or customize your analysis:

1. bdgcmp can be used on *_treat_pileup.bdg and *_control_lambda.bdg or bedGraph files from other resources to calculate the score track.

2. bdgpeakcall can be used on *_treat_pvalue.bdg or the file generated from bdgcmp or bedGraph file from other resources to call peaks with given cutoff, maximum-gap between nearby mergeable peaks and a minimum length of peak. bdgbroadcall works similarly to bdgpeakcall, however, it will output _broad_peaks.bed in BED12 format.

3. Differential calling tool -- bdgdiff, can be used on 4 bedGraph files which are scores between treatment 1 and control 1, treatment 2 and control 2, treatment 1 and treatment 2, treatment 2 and treatment 1. It will output consistent and unique sites according to parameter settings for minimum length, the maximum gap and cutoff.

4. You can combine subcommands to do a step-by-step peak calling. Read detail at MACS2 wikipage