chip-seq-signal-pileup-extension

ChIP-Seq Signal Pileup Extension

Summary

Extend aligned ChIP-Seq reads to their predicted fragment length and generate genome-wide coverage (pileup) tracks. This is a critical intermediate step in peak calling that converts single-end or paired-end reads into continuous signal intensity maps needed for subsequent statistical comparison against background.

When to use

After duplicate filtering and fragment length prediction (d) in ChIP-Seq analysis, when you need to convert discrete read alignments into continuous coverage signal for comparison against control background. Specifically, apply this skill when preparing ChIP pileup tracks before constructing local bias models or performing statistical testing with bdgcmp.

When NOT to use

• When input reads are already in bedgraph or BigWig format (already represent continuous signal, not discrete alignments).
• When analyzing broad histone marks without prior fragment length prediction; use bdgbroadcall instead of narrow peak calling workflow.
• When fragment length d is unknown or invalid (negative, zero, or larger than biological expectation); predictd must succeed first.

Inputs

• Filtered ChIP-Seq BED file (duplicate-filtered reads with columns: chromosome, start, end, name, score, strand)
• Filtered control BED file (duplicate-filtered reads in BED format, optional but recommended)
• Predicted fragment length d in base pairs (integer, e.g. 254)
• Genome size (string code like 'hs' for human, or integer)

Outputs

• ChIP pileup bedgraph file (chromosome, start, end, coverage depth)
• Control pileup bedgraph file (chromosome, start, end, coverage depth)
• d-background bedgraph (control extended to d/2 bp, used for local bias at fragment scale)
• slocal-background bedgraph (control coverage in 1 kb window, used for local bias)
• llocal-background bedgraph (control coverage in 10 kb window, used for local bias)

How to apply

Use macs3 pileup on the filtered ChIP BED file with --extsize set to the predicted fragment length d (e.g., 254 bp for CTCF). This extends each aligned read in both directions to simulate the actual DNA fragment size, creating a bedgraph file where each genomic position is assigned a read count representing local sequencing depth. The extension parameter is critical because it normalizes for the fragment length that was determined in the predictd step, allowing proper comparison with control signal that is similarly extended. Generate separate pileup tracks for both ChIP and control samples; for control, additionally apply the -B flag to generate bidirectional background tracks at multiple scales (d/2, 1 kb, 10 kb) used in local bias calculation. The output bedgraph format preserves base-pair resolution coverage for downstream statistical testing.