normalize-bigwig

Put corrected cut-site tracks on a comparable signal scale using the same background regions for every sample. This is an optional step after atac-correct; use the resulting tracks for downstream scoring or plotting when you need this normalization.

Example command

normalize-bigwig \
  --sample-table project/metadata/samples.tsv \
  --background project/peaks/merged_peaks_filtered.bed \
  --outdir project \
  --method background-scale \
  --stat q95 \
  --target median

Primary inputs

• --sample-table — TSV with sample and condition columns; reuse the table from atac-correct. Tracks are read from {project}/samples/{sample}/atac_correct/.
• --background — shared BED intervals used to calculate comparable background statistics.
• --outdir — project directory represented by {project} below.
• --method — transformation; background-scale multiplies each signal by a shared-target scale factor.
• --stat — within-sample background statistic; the example uses the 95th percentile.
• --target — across-sample target for the selected statistic; the example uses the median.

Main outputs

Path	Meaning
{project}/samples/{sample}/normalize/{sample}_corrected_q95_scaled.bw	Q95-scaled bias-corrected cut-site signal bigWig for one sample.
{project}/normalize_bigwig_qc.tsv	Background statistics, selected statistic, target, and scale factor for every sample.
{project}/normalize_bigwig_manifest.tsv	Sample-to-input/output signal mapping for downstream use.

Check the QC table to see how much each track was scaled. The manifest lists the output paths to pass to the next command. Scaling changes signal magnitude; it does not by itself provide evidence of TF binding.

In custom layout, default outputs use {outdir}/{input_stem}.background_scale_{stat}.bw, plus the two QC tables in {outdir}. background-zscore instead writes standardized signal and uses a method-specific filename suffix.

Download a representative ENCODE normalization QC table. See the Bulk ATAC-seq workflow and the complete normalize-bigwig reference.