DNase-seq

The assay relies on the DNase I endonuclease, which preferentially cleaves DNA in nucleosome-depleted regions where the minor groove is exposed. In tightly packed heterochromatin, the enzyme's activity is sterically hindered. By titrating enzyme concentration, the assay introduces single-strand nicks and double-strand breaks specifically at active regulatory elements, such as promoters and enhancers, generating a library of fragments that map precisely to open chromatin.

Beyond simple peak identification, DNase-seq enables digital genomic footprinting. When a transcription factor binds within an open chromatin region, it physically protects the underlying DNA from DNase I cleavage. This results in a localized dip in sequencing coverage directly over the transcription factor binding site (TFBS). The resulting footprint provides nucleotide-resolution evidence of protein occupancy, distinguishing transient binding from unoccupied accessible DNA.

The experimental workflow involves controlled digestion of intact nuclei, followed by careful size selection of the resulting fragments:

• Small fragments (< 300 bp): Enriched for mononucleosomal and sub-nucleosomal particles, representing true open chromatin.
• Large fragments: Often derived from bulk, undigested chromatin. Gel-based or bead-based size selection is critical to enrich the signal-to-noise ratio, ensuring that sequencing reads concentrate on DNase I hypersensitive sites (DHSs) rather than background genomic DNA.

Computational analysis of DNase-seq data requires rigorous statistical modeling to distinguish true hypersensitive sites from stochastic noise. The Poisson distribution or negative binomial models are typically employed to model read counts against a background expectation. The Irreproducible Discovery Rate (IDR) framework is the gold standard for assessing peak consistency across biological replicates, ensuring that only high-confidence, reproducible DHSs are reported for downstream regulatory annotation.

While both assays map open chromatin, they differ fundamentally in mechanism:

• DNase-seq: Uses direct enzymatic cleavage; requires more starting material and a complex titration step but provides superior footprinting resolution.
• ATAC-seq: Uses hyperactive Tn5 transposase to simultaneously fragment and tag DNA; requires fewer cells and has a simpler protocol but can introduce a strong GC bias and mitochondrial DNA contamination. DNase-seq remains the historical gold standard for regulatory element catalogs like the ENCODE Project.

DNase-seq data serves as a critical training target for sequence-to-activity models. Architectures like DeepSEA and Enformer predict DNase-seq signal directly from raw DNA sequence, learning the complex grammar of chromatin accessibility. These models use DNase-seq profiles as quantitative labels to train convolutional and attention-based networks, enabling the in silico prediction of regulatory impact for non-coding variants identified in genome-wide association studies.

Assay for Transposase-Accessible Chromatin uses a hyperactive Tn5 transposase to simultaneously fragment and tag open chromatin. Compared to DNase-seq, it requires fewer cells and provides faster library preparation.

• Maps nucleosome-free regions at high resolution
• Often used for single-cell chromatin accessibility profiling
• Directly competes with DNase-seq for regulatory element discovery

Chromatin Immunoprecipitation sequencing maps the genome-wide binding locations of specific proteins. While DNase-seq identifies all open regions, ChIP-seq pinpoints exactly which transcription factor or histone modification occupies a site.

• Requires a high-quality, specific antibody
• Complements DNase-seq by identifying the trans-acting factors at accessible loci
• The gold standard for mapping histone modification landscapes

A high-resolution computational analysis applied to DNase-seq data. It detects localized dips in the cleavage signal where a transcription factor physically occludes the DNA from DNase I digestion.

• Reveals precise protein-DNA contact sites at nucleotide resolution
• Distinguishes bound factors from general nucleosome depletion
• Computationally intensive, often using algorithms like Wellington or HINT

The computational process of identifying genomic regions with statistically significant enrichment of mapped reads. For DNase-seq, peaks define DNase I hypersensitive sites (DHSs).

• Algorithms like MACS2 or Hotspot model background noise
• Outputs a set of discrete genomic intervals representing active regulatory DNA
• The foundational step before differential accessibility analysis

The physical property measured by DNase-seq. It defines the degree to which nuclear machinery can interact with DNA. Accessible chromatin is a universal marker for active promoters, enhancers, and insulators.

• A binary state: open (euchromatin) or closed (heterochromatin)
• Highly cell-type-specific, defining cellular identity
• The primary phenotype predicted by deep learning models like DeepSEA

A pioneering deep learning framework that predicts chromatin accessibility (DNase-seq signal), transcription factor binding, and histone marks directly from a 1,000-base-pair DNA sequence.

• Uses a convolutional architecture to learn regulatory motifs de novo
• Demonstrates that sequence context alone is highly predictive of epigenetic state
• Enables in silico mutagenesis to predict the impact of non-coding variants