Version Compatibility

Reference examples tested with: bedtools 2.31+, 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.

bedGraph Handling

"Work with bedGraph signal tracks" → Create, manipulate, and convert bedGraph files for displaying coverage or signal intensity on genome browsers.

• CLI:

  bedtools genomecov -bg

  to generate,

  bedGraphToBigWig

  to convert
• Python:

  pyBigWig

  ,

  pybedtools

bedGraph is a text format for displaying continuous-valued data on genome browsers. Common for coverage, signal intensity, and scores.

bedGraph Format

track type=bedGraph name="Sample" description="Coverage"
chr1    0       100     1.5
chr1    100     200     2.3
chr1    200     300     0.8

Four columns: chrom, start, end, value (0-based, half-open)

Create bedGraph from BAM

Using bedtools genomecov

bedtools genomecov -ibam sample.bam -bg > sample.bedgraph
bedtools genomecov -ibam sample.bam -bg -split > sample.bedgraph
bedtools genomecov -ibam sample.bam -bg -scale 1.5 > sample.scaled.bedgraph

Strand-Specific

bedtools genomecov -ibam sample.bam -bg -strand + > sample.plus.bedgraph
bedtools genomecov -ibam sample.bam -bg -strand - > sample.minus.bedgraph

5' End Coverage (ChIP-seq)

bedtools genomecov -ibam sample.bam -bg -5 > sample.5prime.bedgraph

Normalize by Library Size (CPM)

total_reads=$(samtools view -c -F 260 sample.bam)
scale=$(echo "scale=10; 1000000 / $total_reads" | bc)

bedtools genomecov -ibam sample.bam -bg -scale $scale > sample.cpm.bedgraph

Sort bedGraph

bedGraph must be sorted for conversion to bigWig.

sort -k1,1 -k2,2n sample.bedgraph > sample.sorted.bedgraph
LC_ALL=C sort -k1,1 -k2,2n sample.bedgraph > sample.sorted.bedgraph

Convert bedGraph to bigWig

Using UCSC bedGraphToBigWig

bedGraphToBigWig sample.sorted.bedgraph chrom.sizes sample.bw
fetchChromSizes hg38 > hg38.chrom.sizes
bedGraphToBigWig sample.sorted.bedgraph hg38.chrom.sizes sample.bw

Generate chrom.sizes

samtools faidx reference.fa
cut -f1,2 reference.fa.fai > chrom.sizes
fetchChromSizes hg38 > hg38.chrom.sizes
mysql --user=genome --host=genome-mysql.soe.ucsc.edu -A -e \
    "select chrom, size from hg38.chromInfo" > hg38.chrom.sizes

Clip to Chromosome Boundaries

bedClip sample.bedgraph chrom.sizes sample.clipped.bedgraph
bedGraphToBigWig sample.clipped.bedgraph chrom.sizes sample.bw

Convert bigWig to bedGraph

bigWigToBedGraph sample.bw sample.bedgraph
bigWigToBedGraph sample.bw sample.chr1.bedgraph -chrom=chr1
bigWigToBedGraph sample.bw sample.region.bedgraph -chrom=chr1 -start=1000 -end=2000

Merge bedGraph Files

Using bedtools unionbedg

bedtools unionbedg -i sample1.bedgraph sample2.bedgraph sample3.bedgraph \
    -header -names sample1 sample2 sample3 > merged.bedgraph

Average Across Samples

bedtools unionbedg -i sample1.bedgraph sample2.bedgraph sample3.bedgraph | \
    awk '{sum=0; for(i=4;i<=NF;i++) sum+=$i; print $1,$2,$3,sum/(NF-3)}' OFS='\t' \
    > average.bedgraph

Mathematical Operations

bedtools map for Region Statistics

bedtools map -a regions.bed -b sample.bedgraph -c 4 -o mean > region_means.bed
bedtools map -a regions.bed -b sample.bedgraph -c 4 -o sum > region_sums.bed
bedtools map -a regions.bed -b sample.bedgraph -c 4 -o max > region_max.bed

Subtract Background

bedtools unionbedg -i treatment.bedgraph input.bedgraph | \
    awk '{diff=$4-$5; if(diff<0) diff=0; print $1,$2,$3,diff}' OFS='\t' \
    > subtracted.bedgraph

Log Transform

awk '{print $1,$2,$3,log($4+1)/log(2)}' OFS='\t' sample.bedgraph > sample.log2.bedgraph

Smooth Signal

bedtools slop -i sample.bedgraph -g chrom.sizes -b 50 | \
    bedtools merge -i - -c 4 -o mean > smoothed.bedgraph

Python with pyBigWig

Write bedGraph

import pyBigWig

bw = pyBigWig.open('output.bedgraph', 'w')
bw.addHeader([('chr1', 248956422), ('chr2', 242193529)])

chroms = ['chr1', 'chr1', 'chr1']
starts = [0, 100, 200]
ends = [100, 200, 300]
values = [1.5, 2.3, 0.8]
bw.addEntries(chroms, starts, ends=ends, values=values)
bw.close()

Read bigWig to bedGraph Format

import pyBigWig

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

for chrom, size in bw.chroms().items():
    intervals = bw.intervals(chrom)
    if intervals:
        for start, end, value in intervals:
            print(f'{chrom}\t{start}\t{end}\t{value}')

bw.close()

Convert bigWig Region to bedGraph

import pyBigWig

bw = pyBigWig.open('sample.bw')
intervals = bw.intervals('chr1', 1000000, 2000000)

with open('region.bedgraph', 'w') as f:
    for start, end, value in intervals:
        f.write(f'chr1\t{start}\t{end}\t{value}\n')

bw.close()

deepTools for Normalization

bamCoverage (BAM to bedGraph/bigWig)

bamCoverage -b sample.bam -o sample.bw --normalizeUsing RPKM
bamCoverage -b sample.bam -o sample.bw --normalizeUsing CPM
bamCoverage -b sample.bam -o sample.bw --normalizeUsing BPM
bamCoverage -b sample.bam -o sample.bedgraph --outFileFormat bedgraph --normalizeUsing CPM

bamCompare (Treatment vs Control)

bamCompare -b1 treatment.bam -b2 input.bam -o log2ratio.bw --scaleFactorsMethod readCount
bamCompare -b1 treatment.bam -b2 input.bam -o subtracted.bw --ratio subtract

bigwigCompare

bigwigCompare -b1 treatment.bw -b2 input.bw -o ratio.bw --ratio log2
bigwigCompare -b1 sample1.bw -b2 sample2.bw -o diff.bw --ratio subtract

Filtering and Subsetting

Filter by Value

awk '$4 >= 1.0' sample.bedgraph > high_signal.bedgraph
awk '$4 > 0' sample.bedgraph > nonzero.bedgraph

Extract Regions

bedtools intersect -a sample.bedgraph -b regions.bed > subset.bedgraph

Remove Specific Chromosomes

grep -v "^chrM" sample.bedgraph | grep -v "_random" > filtered.bedgraph
awk '$1 ~ /^chr[0-9XY]+$/' sample.bedgraph > standard_chroms.bedgraph

Aggregate to Bins

Fixed-Size Bins

bedtools makewindows -g chrom.sizes -w 1000 > bins.bed
bedtools map -a bins.bed -b sample.bedgraph -c 4 -o mean > binned.bedgraph

Gene Bodies

bedtools map -a genes.bed -b sample.bedgraph -c 4 -o mean > gene_signal.bed

Quality Control

Check for Overlapping Intervals

bedtools merge -i sample.bedgraph -c 4 -o collapse | \
    awk 'index($4,",") > 0' | head

Verify Sorted Order

sort -c -k1,1 -k2,2n sample.bedgraph && echo "Sorted" || echo "Not sorted"

Check Value Range

awk 'NR==1 {min=$4; max=$4} {if($4<min) min=$4; if($4>max) max=$4}
     END {print "Min:", min, "Max:", max}' sample.bedgraph

Complete Pipeline

Goal: Generate a CPM-normalized bigWig track from a BAM file in a single automated workflow.

Approach: Count total mapped reads with samtools, compute a CPM scale factor, generate a scaled bedGraph with bedtools genomecov, sort and clip to chromosome boundaries, then convert to bigWig format.

#!/bin/bash
BAM=$1
NAME=$(basename $BAM .bam)
CHROM_SIZES=$2

total_reads=$(samtools view -c -F 260 $BAM)
scale=$(echo "scale=10; 1000000 / $total_reads" | bc)

bedtools genomecov -ibam $BAM -bg -scale $scale > ${NAME}.bedgraph

sort -k1,1 -k2,2n ${NAME}.bedgraph > ${NAME}.sorted.bedgraph

bedClip ${NAME}.sorted.bedgraph $CHROM_SIZES ${NAME}.clipped.bedgraph

bedGraphToBigWig ${NAME}.clipped.bedgraph $CHROM_SIZES ${NAME}.bw

rm ${NAME}.bedgraph ${NAME}.sorted.bedgraph ${NAME}.clipped.bedgraph

echo "Created ${NAME}.bw (CPM normalized)"

Track Header for UCSC

echo 'track type=bedGraph name="Sample" description="CPM normalized" visibility=full color=0,0,255 altColor=255,0,0 autoScale=on graphType=bar' > track.bedgraph
cat sample.bedgraph >> track.bedgraph

Related Skills

• coverage-analysis - Generate coverage from alignments
• bigwig-tracks - Work with bigWig format
• chip-seq/chipseq-visualization - Visualize signal tracks
• alignment-files/sam-bam-basics - BAM file processing