chip-seq-read-alignment-filtering

ChIP-Seq read alignment filtering

Summary

Filter redundant reads from ChIP-Seq alignment files and perform duplicate-aware read counting prior to peak calling. This is a critical initial step in MACS3 callpeak analysis that removes PCR artifacts and establishes the true library complexity, which affects all downstream fragment length prediction and background modeling.

When to use

When beginning peak calling on ChIP-Seq data: you have raw single-end or paired-end BED/BEDPE alignment files for both ChIP and control samples and need to remove duplicate reads before predicting fragment length and building local bias models. This step is obligatory before macs3 predictd and macs3 pileup.

When NOT to use

• Input is already a deduplicated or UMI-collapsed alignment file (would lose information on true duplicate complexity).
• Analysis goal is to preserve all reads for coverage visualization without statistical peak calling (use macs3 pileup directly instead).
• Data is single-cell ChIP-Seq or other ultra-sparse ChIP where aggressive duplicate filtering would remove real signal at low coverage sites.

Inputs

• ChIP sample alignment file in BED format (e.g., CTCF_ChIP_200K.bed.gz)
• Control sample alignment file in BED format (e.g., CTCF_Control_200K.bed.gz)

Outputs

• Filtered ChIP BED file with duplicates removed
• Filtered control BED file with duplicates removed
• Read count statistics (number of reads retained after duplicate filtering)

How to apply

Apply macs3 filterdup separately to both ChIP and control BED files using --keep-dup=1 to retain one copy of each duplicate read cluster (or --keep-dup=all to retain all duplicates, depending on sequencing depth and library complexity expectations). The command reads genomic coordinates from the alignment file, identifies reads mapping to identical positions, and outputs filtered BED with duplicate read counts recorded. Record the final read counts for each sample (e.g., ChIP: 199,583; Control: 199,867) as these are used to compute genome-wide background (control_reads × fragment_length / genome_size) and the ChIP-to-control scaling ratio in later steps. The duplicate filtering rate and absolute read counts are quality metrics that should be assessed before proceeding to fragment length prediction.