Summary De novo motif discovery and known motif enrichment analysis using HOMER and MEME-ChIP. Identify transcription factor binding motifs in ChIP-seq, ATAC-seq, or other genomic peak data. Use when finding enriched DNA motifs in peak sequences.
Version Compatibility
Reference examples tested with: BioPython 1.83+, bedtools 2.31+, matplotlib 3.8+, pandas 2.2+
Before using code patterns, verify installed versions match. If versions differ:
Python: pip show <package> then help(module.function) to check signatures
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.
Motif Analysis
"Find enriched motifs in my ChIP-seq peaks" → Discover de novo DNA-binding motifs and test for known TF motif enrichment in peak sequences.
CLI: findMotifsGenome.pl peaks.bed hg38 output/ (HOMER), meme-chip -db JASPAR peaks.fa (MEME)
Identify DNA sequence motifs enriched in ChIP-seq or ATAC-seq peaks to discover transcription factor binding sites.
Tool Comparison
Tool Strengths Use Case HOMER Fast, comprehensive, built-in databases General motif analysis MEME-ChIP Multiple algorithms, web interface Publication-quality MEME De novo discovery only Simple discovery FIMO Known motif scanning Genome-wide scanning
HOMER
Installation
conda install -c bioconda homer
# Configure genome (required once)
perl /path/to/homer/configureHomer.pl -install hg38
perl /path/to/homer/configureHomer.pl -install mm10
De Novo Motif Discovery Goal: Discover enriched DNA-binding motifs directly from ChIP-seq peak sequences.
Approach: Run findMotifsGenome.pl on a peak BED file with a specified fragment size, optionally providing background regions and target motif lengths.
# Basic motif finding
findMotifsGenome.pl peaks.bed hg38 output_dir/ -size 200
# With background regions
findMotifsGenome.pl peaks.bed hg38 output_dir/ -size 200 -bg background.bed
# Specify motif lengths to search
findMotifsGenome.pl peaks.bed hg38 output_dir/ -size 200 -len 8,10,12
Key Options Option Description -size <#>Fragment size for analysis (default 200) -size givenUse actual peak sizes -bg <file>Background regions (BED) -len <#,#,...>Motif lengths to search -maskMask repeats -p <#>Number of CPUs -S <#>Number of motifs to find (default 25) -mis <#>Mismatches allowed (default 2) -noweightDon't adjust for GC content
Output Files output_dir/
├── homerResults.html # Main results page
├── knownResults.html # Known motif enrichment
├── homerMotifs.all.motifs # All discovered motifs
├── knownResults.txt # Known motif statistics
└── motif1.motif # Individual motif files
Known Motif Enrichment Only # Skip de novo, only check known motifs
findMotifsGenome.pl peaks.bed hg38 output_dir/ -size 200 -nomotif
Scan for Specific Motifs # Find instances of motif in peaks
annotatePeaks.pl peaks.bed hg38 -m motif.motif > annotated.txt
# Scan genome for motif occurrences
scanMotifGenomeWide.pl motif.motif hg38 > motif_sites.bed
Motif Comparison # Compare discovered motifs to known database
compareMotifs.pl motifs.motif output_dir/ -known
Create Custom Motif # From consensus sequence
seq2profile.pl CACGTG 4 > MYC.motif
# From aligned sequences
cat aligned_seqs.txt | alignAndConvert.pl - > custom.motif
MEME Suite
Installation conda install -c bioconda meme
Extract Sequences from Peaks # Get FASTA sequences under peaks
bedtools getfasta -fi genome.fa -bed peaks.bed -fo peaks.fa
# Center peaks and resize
bedtools slop -i peaks.bed -g genome.sizes -b 100 | \
bedtools getfasta -fi genome.fa -bed - -fo peaks_centered.fa
MEME (De Novo Discovery) # Basic de novo discovery
meme peaks.fa -dna -oc meme_output -mod zoops -nmotifs 10 -minw 6 -maxw 20
# With Markov background
fasta-get-markov peaks.fa > background.model
meme peaks.fa -dna -oc meme_output -bfile background.model -mod zoops -nmotifs 10
MEME Options Option Description -mod zoopsZero or one per sequence (default for ChIP) -mod oopsExactly one per sequence -mod anrAny number of repeats -nmotifs <#>Number of motifs to find -minw <#>Minimum motif width -maxw <#>Maximum motif width -revcompSearch both strands -bfile <file>Background model file
MEME-ChIP (Comprehensive Pipeline) Goal: Run a comprehensive motif analysis pipeline combining de novo discovery, central enrichment testing, and database comparison.
Approach: Provide peak FASTA sequences and a motif database to MEME-ChIP, which runs MEME, DREME, CentriMo, TOMTOM, and FIMO in a single invocation.
# All-in-one ChIP-seq motif analysis
meme-chip -oc meme_chip_output -db motif_database.meme peaks.fa
MEME - De novo discovery (central enrichment)
DREME - Short motif discovery
CentriMo - Central enrichment analysis
TOMTOM - Compare to known motifs
FIMO - Find motif instances
DREME (Short Motifs) # Find short enriched motifs
dreme -oc dreme_output -p peaks.fa -n background.fa
CentriMo (Central Enrichment) # Test for central enrichment of known motifs
centrimo -oc centrimo_output peaks.fa motif_database.meme
TOMTOM (Motif Comparison) # Compare discovered motifs to database
tomtom -oc tomtom_output discovered.meme database.meme
FIMO (Motif Scanning) # Scan sequences for motif matches
fimo --oc fimo_output motif.meme sequences.fa
# Scan genome
fimo --oc fimo_output --max-stored-scores 1000000 motif.meme genome.fa
Motif Databases
HOMER Built-in # List available motif sets
ls /path/to/homer/data/knownTFs/
# Vertebrate, known motifs (default)
findMotifsGenome.pl peaks.bed hg38 output/ -mknown vertebrates/known.motifs
JASPAR # Download JASPAR motifs
wget https://jaspar.genereg.net/download/data/2024/CORE/JASPAR2024_CORE_vertebrates_non-redundant_pfms_meme.txt
# Use with MEME suite
meme-chip -db JASPAR2024_CORE_vertebrates_non-redundant_pfms_meme.txt peaks.fa
HOCOMOCO # Download HOCOMOCO
wget https://hocomoco11.autosome.org/final_bundle/hocomoco11/core/HUMAN/mono/HOCOMOCOv11_core_HUMAN_mono_meme_format.meme
# Use with MEME suite
tomtom discovered.meme HOCOMOCOv11_core_HUMAN_mono_meme_format.meme
Python: Parse HOMER Results import pandas as pd
def parse_homer_known(results_file):
'''Parse HOMER knownResults.txt.'''
df = pd.read_csv(results_file, sep='\t')
df.columns = ['Motif', 'Consensus', 'P-value', 'Log P-value',
'q-value', 'Targets', 'Target%', 'Background', 'Background%']
df['P-value'] = df['P-value'].astype(float)
return df.sort_values('P-value')
known = parse_homer_known('output_dir/knownResults.txt')
print(known[['Motif', 'P-value', 'Target%']].head(20))
Python: Parse MEME Results from Bio import motifs
def parse_meme_file(meme_file):
'''Parse MEME output file.'''
with open(meme_file) as f:
record = motifs.parse(f, 'meme')
return record
record = parse_meme_file('meme_output/meme.txt')
for m in record:
print(f'{m.name}: {m.consensus}')
print(m.counts)
Complete Workflows
ChIP-seq Motif Analysis Goal: Run a complete motif analysis workflow combining HOMER and MEME-ChIP on ChIP-seq peaks.
Approach: Run HOMER findMotifsGenome.pl for fast de novo and known motif discovery, then extract centered peak sequences and run MEME-ChIP for a complementary analysis.
#!/bin/bash
set -euo pipefail
PEAKS=$1 # narrowPeak or BED file
GENOME=$2 # hg38, mm10, etc.
OUTDIR=$3
mkdir -p $OUTDIR
# HOMER analysis
echo "Running HOMER..."
findMotifsGenome.pl $PEAKS $GENOME ${OUTDIR}/homer \
-size 200 -p 8 -mask
# Extract sequences for MEME
echo "Extracting sequences..."
bedtools slop -i $PEAKS -g ${GENOME}.chrom.sizes -b 0 | \
awk 'BEGIN{OFS="\t"} {center=int(($2+$3)/2); print $1,center-100,center+100}' | \
bedtools getfasta -fi ${GENOME}.fa -bed - -fo ${OUTDIR}/peaks.fa
# MEME-ChIP analysis
echo "Running MEME-ChIP..."
meme-chip -oc ${OUTDIR}/meme_chip \
-db /path/to/JASPAR.meme \
${OUTDIR}/peaks.fa
echo "Done. Results in ${OUTDIR}/"
ATAC-seq Footprint Motifs # Analyze motifs in footprint regions
findMotifsGenome.pl footprints.bed hg38 footprint_motifs/ \
-size given -mask -p 8
# Compare to accessible regions background
findMotifsGenome.pl footprints.bed hg38 footprint_motifs/ \
-size given -bg accessible_peaks.bed -mask -p 8
Visualization
HOMER Logo # Generate sequence logo
motif2Logo.pl motif.motif > logo.eps
Plot with Python import logomaker
import pandas as pd
import matplotlib.pyplot as plt
def plot_motif(pwm_file):
'''Plot sequence logo from HOMER PWM.'''
pwm = pd.read_csv(pwm_file, sep='\t', skiprows=1, header=None)
pwm.columns = ['A', 'C', 'G', 'T']
logo = logomaker.Logo(pwm, shade_below=0.5, fade_below=0.5)
plt.show()
Quality Metrics Metric Good Concerning P-value < 1e-10 > 1e-5 Target % > 20% < 5% Background % < Target/2 Similar to Target Bit score > 10 < 5
Common Issues
No Significant Motifs
Check peak quality (too few peaks?)
Try different peak sizes (-size)
Ensure genome build matches
Check for repeat masking issues
Too Many Motifs
Increase significance threshold
Use -S to limit number of motifs
Filter by target percentage
Wrong Background
Use matched GC content background
Consider using input/control peaks
Try shuffled sequences
Related Skills
peak-calling - Generate input peaks
peak-annotation - Annotate peaks with genes
atac-seq/footprinting - TF footprint analysis
genome-intervals - BED file operations
