MAmotif is used to compare two ChIP-seq samples of the same protein from different cell types or conditions (e.g. Mutant vs Wild-type) and identify transcriptional factors (TFs) associated with the cell-type biased binding of this protein as its co-factors, by using TF binding information obtained from motif analysis (or from other ChIP-seq data).
MAmotif automatically combines MAnorm model to perform quantitative comparison on given ChIP-seq samples together with Motif-Scan toolkit to scan ChIP-seq peaks for TF binding motifs, and uses a systematic integrative analysis to search for TFs whose binding sites are significantly associated with the cell-type biased peaks between two ChIP-seq samples.
When applying to ChIP-seq data of histone marks of regulatory elements (such as H3K4me3 for active promoters and H3K9/27ac for active promoter/enhancers), or DNase/ATAC-seq data, MAmotif can be used to detect cell-type specific regulators.
To see the full documentation of MAmotif, please refer to: http://mamotif.readthedocs.io/en/latest/
The latest release of MAmotif is available at PyPI:
$ pip install mamotif
Or you can install MAmotif via conda:
$ conda install -c bioconda mamotif
MAmotif uses setuptools for installation from source code. The source code of MAmotif is hosted on GitHub: https://github.com/shao-lab/MAmotif
You can clone the repo and execute the following command under source directory:
$ python setup.py install
WIP!
You need to build some prerequisites before running MAmotif:
Preprocess sequences and genome-wide nucleotide frequency for the corresponding genome assembly.
$ genomecompile [-h] [-v] -G hg19.fa -o hg19_genome
Note: You only need to run this command once for each genome
Note: MAmotif provides some preprocessed motif PWM files under data/motif of the MotifScan package.
Build motif PWM/motif-score cutoff for custom motifs that are not included in our pre-complied motif collection:
$ motifcompile [-h] [-v] –M motif_pwm_demo.txt –g hg19_genome -o hg19_motif
$ mamotif --p1 sample1_peaks.bed --p2 sample2_peaks.bed --r1 sample1_reads.bed --r2 sample2_reads.bed -g hg19_genome –m hg19_motif_p1e-4.txt -o sample1_vs_sample2
Note: Using -h/--help for the details of all arguments.
After finished running MAmotif, all output files will be written to the directory you specified with "-o" argument.
1.Motif Name 2.Target Number: Number of motif-present peaks 3.Average of Target M-value: Average M-value of motif-present peaks 4.Deviation of Target M-value: M-value Std of motif-present peaks 5.Non-target Number: Number of motif-absent peaks 6.Average of Non-target M-value: Average M-value of motif-absent peaks 7.Deviation of Non-target M-value: M-value Std of motif-absent peaks 8.T-test Statistics: T-Statistics for M-values of motif-present peaks against motif-absent peaks 9.T-test P-value: Right-tailed P-value of T-test 10.T-test P-value By Benjamin correction 11.RanSum-test Statistics 12.RankSum-test P-value 13.RankSum-test P-value By Benjamin correction 14.Maximal P-value: Maximal corrected P-value of T-test and RankSum-test
MAmotif will invoke MAnorm and output the normalization results and MA-plot for samples under comparison.
MAmotif will also output tables to summarize the enrichment of motifs and the motif target number and motif-score of each peak region.
If you specified "-s" with MAmotif, it will also output the genome coordinates of every motif target site.
Here we provide a step-by-step instruction on how to use MAmotif to find candidate cell-type specific regulators associated with certain histone modifications.
We take the H3K4me3 analysis between adult and fetal ProES in MAmotif paper as an example:
Install MAmotif:
$pip install mamotif $conda install -c bioconda mamotif
Download all data MAmotif needs:
$mkdir MAmotif_demo $cd MAmotif_demo $wget ftp://ftp.ncbi.nlm.nih.gov/geo/samples/GSM908nnn/GSM908038/suppl/GSM908038_H3K4me3-F_peaks.bed.gz $wget ftp://ftp.ncbi.nlm.nih.gov/geo/samples/GSM908nnn/GSM908039/suppl/GSM908039_H3K4me3-A_peaks.bed.gz $wget ftp://ftp.ncbi.nlm.nih.gov/geo/samples/GSM908nnn/GSM908038/suppl/GSM908038_H3K4me3-F.bed.gz $wget ftp://ftp.ncbi.nlm.nih.gov/geo/samples/GSM908nnn/GSM908039/suppl/GSM908039_H3K4me3-A.bed.gz $gzip -d *gz Remove the header line and ribosomal reads (You do not need to do this for modern ChIP-seq mapping softwares) $sed -i '1d' GSM908038_H3K4me3-F.bed $sed -i '1d' GSM908039_H3K4me3-A.bed $sed -i '8986927,$d' GSM908039_H3K4me3-F.bed $sed -i '14916308,$d' GSM908039_H3K4me3-A.bed Substitute space into tab for bed files (You do not need to do this for your own bed files are tab-separated) $sed -i "s/ /\t/g" GSM908038_H3K4me3-F.bed $sed -i "s/ /\t/g" GSM908039_H3K4me3-A.bed
Build for genome sequences:
$mkdir genome $cd genome $wget http://hgdownload.cse.ucsc.edu/goldenPath/hg18/bigZips/chromFa.zip $unzip chromFa.zip $cat *fa > hg18.fa $genomecompile -G hg18.fa -o hg18 $cd ..
Build for motif PWM (Optional)
The motif matrix file which containing the motif score cutoff is already packaged under /data directory under MotifScan package.
If you want you compile for your custom motifs, please run the following commands:
$mkdir motif $cd motif $wget http://jaspar2016.genereg.net/html/DOWNLOAD/JASPAR_CORE/pfm/nonredundant.tar.gz $tar -xzvf nonredundant.tar.gz $motifcompile -M nonredundant/pfm_vertebrates.txt -g ../genome/hg18 -o hg18_jaspar2016_nonredundant_vertebrates $cd ..
Run MAmotif:
$mamotif --p1 GSM908039_H3K4me3-A_peaks.bed --p2 GSM908038_H3K4me3-F_peaks.bed --r1 GSM908039_H3K4me3-A.bed --r2 GSM908038_H3K4me3-F.bed -g genome/hg18 -m motif/hg18_jaspar2016_nonredundant_vertebrates_1e-4.txt -o AvsF_H3K4me3_MAmotif
Check the output of MAmotif