Version Compatibility

Reference examples tested with: bedtools 2.31+, numpy 1.26+, pandas 2.2+, samtools 1.19+

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.

BigWig Tracks

"Read and create BigWig files" → Access indexed binary signal tracks for efficient region queries and genome browser display.

• Python: pyBigWig.open('file.bw') (pyBigWig)
• CLI: bigWigToBedGraph, bedGraphToBigWig (UCSC tools)

BigWig is an indexed binary format for continuous genomic data. Efficient for genome browsers and programmatic access.

Why BigWig?

Format	Size	Random Access	Browser Support
bedGraph	Large	No	Limited
bigWig	~10x smaller	Yes (indexed)	Excellent

Convert bedGraph to bigWig (CLI)

Installation

# UCSC tools
conda install -c bioconda ucsc-bedgraphtobigwig ucsc-bigwigtobedgraph

# Or download directly
wget http://hgdownload.soe.ucsc.edu/admin/exe/linux.x86_64/bedGraphToBigWig
chmod +x bedGraphToBigWig

Basic Conversion

# Sort bedGraph first (required)
sort -k1,1 -k2,2n coverage.bedGraph > coverage.sorted.bedGraph

# Convert to bigWig
bedGraphToBigWig coverage.sorted.bedGraph chrom.sizes output.bw

# chrom.sizes format: chr<TAB>size
# chr1	248956422
# chr2	242193529

Get Chromosome Sizes

# From FASTA index
cut -f1,2 reference.fa.fai > chrom.sizes

# Download from UCSC
wget https://hgdownload.soe.ucsc.edu/goldenPath/hg38/bigZips/hg38.chrom.sizes

# From BAM header
samtools view -H alignments.bam | grep @SQ | sed 's/@SQ\tSN:\|LN://g' > chrom.sizes

Full Workflow

# Generate bedGraph from BAM
bedtools genomecov -ibam alignments.bam -bg > coverage.bedGraph

# Sort bedGraph
sort -k1,1 -k2,2n coverage.bedGraph > coverage.sorted.bedGraph

# Convert to bigWig
bedGraphToBigWig coverage.sorted.bedGraph hg38.chrom.sizes coverage.bw

# Clean up intermediate files
rm coverage.bedGraph coverage.sorted.bedGraph

Read BigWig with pyBigWig (Python)

Installation

pip install pyBigWig

Open and Inspect

import pyBigWig

# Open file
bw = pyBigWig.open('coverage.bw')

# File info
print(f'Chromosomes: {bw.chroms()}')
print(f'Header: {bw.header()}')

# Check if file is bigWig (not bigBed)
print(f'Is bigWig: {bw.isBigWig()}')

# Close when done
bw.close()

Extract Values

import pyBigWig

bw = pyBigWig.open('coverage.bw')

# Get values for a region (returns numpy array)
values = bw.values('chr1', 1000000, 1001000)
print(f'Mean: {values.mean():.2f}')
print(f'Max: {values.max():.2f}')

# Get specific intervals with values
intervals = bw.intervals('chr1', 1000000, 1001000)
# Returns: [(start, end, value), ...]
for start, end, val in intervals:
    print(f'{start}-{end}: {val}')

# Statistics for region
stats = bw.stats('chr1', 1000000, 1001000, type='mean')
print(f'Mean coverage: {stats[0]:.2f}')

# Available stat types: mean, min, max, coverage, std, sum
max_val = bw.stats('chr1', 1000000, 1001000, type='max')
coverage = bw.stats('chr1', 1000000, 1001000, type='coverage')

bw.close()

Binned Statistics

import pyBigWig

bw = pyBigWig.open('coverage.bw')

# Get mean values in 100bp bins across region
region_start, region_end = 1000000, 2000000
n_bins = 1000  # 100bp bins

binned = bw.stats('chr1', region_start, region_end, type='mean', nBins=n_bins)
# Returns list of n_bins values

bw.close()

Extract for BED Regions

Goal: Extract mean bigWig signal values for a set of genomic regions defined in a BED file.

Approach: Open the bigWig file with pyBigWig, iterate through BED intervals, query the mean signal per region, and collect results into a pandas DataFrame.

import pyBigWig
import pybedtools

bw = pyBigWig.open('coverage.bw')
bed = pybedtools.BedTool('regions.bed')

# Get mean signal per region
results = []
for interval in bed:
    chrom, start, end = interval.chrom, interval.start, interval.end
    mean_signal = bw.stats(chrom, start, end, type='mean')[0]
    results.append({
        'chrom': chrom,
        'start': start,
        'end': end,
        'name': interval.name,
        'signal': mean_signal if mean_signal else 0
    })

bw.close()

# Convert to DataFrame
import pandas as pd
df = pd.DataFrame(results)
print(df)

Create BigWig with pyBigWig

import pyBigWig

# Create new bigWig
bw = pyBigWig.open('output.bw', 'w')

# Add header (chromosome sizes)
bw.addHeader([('chr1', 248956422), ('chr2', 242193529)])

# Add entries (must be sorted by position)
# Method 1: Individual entries
bw.addEntries(['chr1', 'chr1'], [0, 100], ends=[100, 200], values=[1.5, 2.3])

# Method 2: Chromosome at a time (more efficient)
bw.addEntries('chr1', [0, 100, 200], ends=[100, 200, 300], values=[1.5, 2.3, 3.1])

# Method 3: Fixed-width spans (most efficient for dense data)
bw.addEntries('chr2', 0, values=[1.0, 2.0, 3.0, 4.0], span=100, step=100)
# Creates: chr2:0-100=1.0, chr2:100-200=2.0, chr2:200-300=3.0, chr2:300-400=4.0

bw.close()

deepTools for BigWig Operations

Installation

conda install -c bioconda deeptools

Generate Normalized BigWig from BAM

# RPKM normalization
bamCoverage -b alignments.bam -o coverage.bw --normalizeUsing RPKM

# CPM normalization
bamCoverage -b alignments.bam -o coverage.bw --normalizeUsing CPM

# BPM (bins per million) - like TPM for ChIP-seq
bamCoverage -b alignments.bam -o coverage.bw --normalizeUsing BPM

# With bin size and smoothing
bamCoverage -b alignments.bam -o coverage.bw \
    --binSize 10 \
    --normalizeUsing CPM \
    --smoothLength 30

# Extend reads to fragment length
bamCoverage -b alignments.bam -o coverage.bw \
    --extendReads 200 \
    --normalizeUsing CPM

Compare BigWig Files

# Log2 ratio of two bigWig files
bigwigCompare -b1 treatment.bw -b2 control.bw -o log2ratio.bw --ratio log2

# Subtract
bigwigCompare -b1 treatment.bw -b2 control.bw -o diff.bw --ratio subtract

# Mean of multiple files
bigwigAverage -b file1.bw file2.bw file3.bw -o average.bw

Summarize BigWig Over Regions

# Matrix for heatmap (signal around regions)
computeMatrix reference-point -S signal.bw -R regions.bed \
    -b 2000 -a 2000 -o matrix.gz

# Plot heatmap
plotHeatmap -m matrix.gz -o heatmap.png

# Summary statistics per region
multiBigwigSummary BED-file -b sample1.bw sample2.bw -o results.npz --BED regions.bed

Convert BigWig to bedGraph

# Using UCSC tool
bigWigToBedGraph input.bw output.bedGraph

# Extract specific region
bigWigToBedGraph input.bw output.bedGraph -chrom=chr1 -start=1000000 -end=2000000

Common Patterns

ChIP-seq Signal Track

# Generate normalized track
bamCoverage -b chip.bam -o chip.bw \
    --normalizeUsing CPM \
    --extendReads 200 \
    --binSize 10

# Generate input-subtracted track
bigwigCompare -b1 chip.bw -b2 input.bw -o chip_minus_input.bw --ratio subtract

RNA-seq Coverage

# Strand-specific coverage
bamCoverage -b rnaseq.bam -o forward.bw --filterRNAstrand forward
bamCoverage -b rnaseq.bam -o reverse.bw --filterRNAstrand reverse

Extract Signal for Analysis

import pyBigWig
import pandas as pd

def extract_signal(bw_path, bed_path, stat='mean'):
    '''Extract bigWig signal for BED regions.'''
    import pybedtools
    bw = pyBigWig.open(bw_path)
    bed = pybedtools.BedTool(bed_path)

    results = []
    for interval in bed:
        val = bw.stats(interval.chrom, interval.start, interval.end, type=stat)[0]
        results.append({
            'chrom': interval.chrom,
            'start': interval.start,
            'end': interval.end,
            'name': interval.name if interval.name else '.',
            'signal': val if val is not None else 0
        })

    bw.close()
    return pd.DataFrame(results)

# Usage
df = extract_signal('coverage.bw', 'peaks.bed', stat='mean')
print(df)

Related Skills

• coverage-analysis - Generate bedGraph input
• chip-seq/chipseq-visualization - ChIP-seq signal tracks
• alignment-files/bam-statistics - BAM to coverage
• interval-arithmetic - Region operations