Strand Cross-Correlation

The ratio of the cross-correlation value at the estimated fragment length peak to the cross-correlation value at the background minimum. NSC quantifies the enrichment of signal at the fragment length relative to the noise floor.

• Calculation: NSC = CC(fragment_length) / CC(minimum)
• Interpretation: Higher values indicate stronger ChIP enrichment
• Quality Threshold: NSC > 1.05 typically indicates successful enrichment
• Failure Mode: NSC ≈ 1.0 suggests no detectable binding signal, often due to poor antibody quality or insufficient sequencing depth

The ratio of the fragment-length cross-correlation to the read-length cross-correlation peak. RSC specifically measures the signal attributable to true ChIP enrichment versus the artificial peak caused by the read length itself.

• Calculation: RSC = CC(fragment_length) / CC(read_length)
• Read-Length Peak: An artifact at exactly the sequencing read length caused by single reads spanning both strands of a short fragment
• Interpretation: RSC > 0.8 indicates strong signal-to-noise ratio
• ENCODE Standard: RSC > 1.0 for point-source factors (e.g., transcription factors); RSC > 0.8 for broad marks (e.g., histone modifications)

The shift distance at which the cross-correlation profile reaches its maximum value after excluding the read-length artifact. This distance corresponds to the predominant fragment length in the ChIP-seq library.

• Derivation: argmax(CC(shift)) for shift > read_length
• Typical Range: 100–300 base pairs for standard ChIP-seq
• Utility: Used to shift reads during peak calling for strand-specific alignment correction
• Diagnostic Value: Unexpectedly short estimates (< 80 bp) may indicate over-sonication; unexpectedly long estimates (> 500 bp) suggest incomplete fragmentation

The ratio of the cross-correlation value at the read-length peak to the value at the background minimum. This metric quantifies the severity of the read-length artifact and is used to diagnose library preparation issues.

• Cause: Arises when a single sequencing read spans an entire short fragment, generating apparent strand overlap at exactly the read length
• Interpretation: Excessively high phantom peaks indicate a high proportion of ultra-short fragments
• Mitigation: Size selection during library preparation to remove adapter-dimers and sub-nucleosomal fragments
• Relationship to RSC: RSC = NSC / Phantom Peak Ratio

The lowest cross-correlation value observed in the profile, typically occurring at shift distances between the read-length peak and the fragment-length peak. This value represents the baseline noise level of the library.

• Location: Usually found at shift distances of 1.5× to 2× the read length
• Interpretation: High background minima indicate elevated non-specific binding or poor library complexity
• Use as Denominator: Serves as the normalization factor for NSC calculation
• Diagnostic: Abnormally high background suggests excessive PCR duplication or insufficient washing during immunoprecipitation

The shape and amplitude of the cross-correlation profile provide indirect measures of library complexity—the number of unique, non-duplicate fragments in the sequencing library.

• Peak Sharpness: A narrow, well-defined fragment-length peak indicates a tight fragment size distribution and high library quality
• Peak Amplitude: Higher absolute cross-correlation values at the fragment length correlate with greater unique fragment diversity
• Non-Redundant Fraction (NRF): While not directly derived from cross-correlation, NRF is often reported alongside NSC and RSC in ENCODE quality reports
• PCR Bottleneck Detection: Broad, low-amplitude peaks with high background suggest low complexity libraries dominated by PCR duplicates