chromatin-accessibility-binding-status-classification

chromatin-accessibility-binding-status-classification

Summary

Classify transcription factor binding sites as bound or unbound by analyzing the spatial distribution of Tn5 insertion depletion patterns (footprints) in ATAC-seq data. This skill leverages the characteristic nucleosome-positioning signal around protein-bound motifs to distinguish occupied from vacant regulatory sites.

When to use

You have ATAC-seq BAM files aligned to a reference genome, a set of transcription factor motif locations (BED format), and you need to determine which motifs are actually occupied by proteins in your cell type or condition. Use this skill when chromatin accessibility alone is insufficient—you require binding status labels for downstream differential binding analysis, network inference, or kinetic studies.

When NOT to use

• Input is single-cell ATAC-seq without pseudobulk aggregation by cell type or condition—individual cell resolution lacks sufficient read depth for reliable footprint detection; use the SC-Framework to generate pseudobulk BAM files first.
• Tn5 insertion bias has not been corrected—biased cutsites can mimic or obscure true footprints; apply ATACorrect before classification.
• You only have summary peak calls (BED) without aligned reads—footprint analysis requires base-pair-resolution insertion positions from BAM files.

Inputs

• ATAC-seq BAM file (aligned reads with insertion coordinates)
• Reference genome FASTA
• Transcription factor motif coordinates (BED format with motif centers)
• ATAC-seq peaks or chromatin accessibility signal (optional; for thresholding accessible sites)

Outputs

• Bound/unbound classification labels per motif site
• Footprint score matrix (insertion counts by position bin and site class)
• Aggregated insertion profiles (BED and BigWig or heatmap visualizations)
• Statistical summaries of positional insertion distributions per class

How to apply

Extract Tn5 insertion coordinates from aligned ATAC-seq reads in fixed-width windows (e.g., ±100 bp) centered on each motif site. Aggregate insertion counts per genomic position bin across all sites within each candidate class. Compute positional distribution statistics (mean, standard deviation, or normalized footprint scores) to quantify the characteristic depletion of insertions around protein-bound motifs—the 'footprint' signal that distinguishes bound from unbound sites. Sites exhibiting strong, focal depletion patterns are classified as bound; those with flat or noisy insertion profiles are classified as unbound. The classification threshold is typically determined by comparing observed insertion profiles to background distributions or by fitting score cutoffs empirically from known positive/negative controls. Visualization as aggregated insertion heatmaps or line plots confirms the expected nucleosome positioning around bound sites.