narrow-peak-calling-score-threshold

narrow-peak-calling-score-threshold

Summary

Call narrow peaks from a q-value bedgraph track by applying a statistical significance cutoff and enforcing minimum peak length and gap constraints. This final step converts continuous score data into discrete peak regions for ChIP-Seq analysis.

When to use

After generating a q-value bedgraph track from ChIP-Seq pileup versus local lambda comparison, and you need to identify statistically significant narrow peaks with defined boundaries. Apply this skill when your input is a bedgraph file of q-values (or p-values) and you want discrete narrowPeak format output with regions that exceed a user-specified significance threshold.

When NOT to use

• Input is broad histone mark data (e.g., H3K27me3, H3K4me1) — use bdgbroadcall instead of bdgpeakcall
• Input bedgraph contains p-values not converted to -log10 scale — ensure proper transformation before applying cutoff
• Fragment length d is unavailable or unreliable (e.g., bimodal or very short ChIP library) — revisit macs3 predictd output or adjust d manually based on expected mark width

Inputs

• bedgraph file of q-value scores (output from macs3 bdgcmp with -m qpois)
• predicted fragment length d (from macs3 predictd)
• significance threshold (q-value cutoff, typically 0.05)

Outputs

• narrowPeak format file with peak coordinates, summit, and -log10(q-value) scores

How to apply

Use macs3 bdgpeakcall to identify contiguous regions in the q-value track that exceed a cutoff threshold (converted from q-value to -log10 scale; e.g., q-value 0.05 = cutoff 1.301). Enforce a minimum peak length equal to the predicted fragment length d (e.g., 245 bp) to filter out spurious short fragments, and set a gap parameter to the read length (typically 100 bp) to merge peaks separated by small intervals. The function outputs narrowPeak format, which includes peak coordinates, summit position, and -log10(q-value) score for each called peak. Rationale: the fragment length constraint ensures peaks are wide enough to represent true ChIP enrichment, while the gap parameter prevents artificial fragmentation of continuous enriched regions.