ATAC-seq is a method for mapping accessible regions of chromatin across a genome. It uses an engineered Tn5 transposase to cut exposed DNA and attach sequencing adapters, allowing the resulting fragments to be identified by high-throughput DNA sequencing. Regions with many recovered fragments are interpreted as relatively accessible under the assay conditions. Introduced in 2013, ATAC-seq became widely used to investigate gene regulation because it combines a comparatively short laboratory workflow with low sample requirements and compatibility with single-cell analysis.
Biological basis
In eukaryotes, DNA is packaged with histone proteins into nucleosomes and higher-order chromatin structures. This packaging influences whether regulatory proteins can interact with particular DNA sequences. Accessible regions frequently include promoters, enhancers, insulators, and other regulatory elements.
Accessibility is related to, but distinct from, gene expression. An accessible promoter or enhancer may be active, poised for later activation, or accessible without contributing detectably to transcription in the measured condition. Conversely, gene regulation also depends on protein binding, chemical modifications, and interactions between distant genomic regions that ATAC-seq does not directly measure.
The method therefore provides a map of physical accessibility rather than a direct inventory of active genes or bound regulatory proteins. Its signal is shaped both by biological chromatin organization and by the biochemical preferences of the assay.
Molecular principle
The central reagent is a hyperactive form of Tn5 transposase, an enzyme derived from a bacterial transposition system. Tn5 is supplied as a complex with adapter-containing DNA molecules. When it encounters accessible DNA, it introduces staggered cuts and transfers adapter sequences to the cut sites.
This combined fragmentation and adapter-tagging reaction is called tagmentation. DNA protected by nucleosomes or other bound proteins is generally less accessible to insertion, although protection is not absolute.
After tagmentation, the tagged fragments are amplified by polymerase chain reaction to generate a sequencing library. Paired-end sequencing identifies both ends of each recovered fragment, providing information about its genomic position and length.
ATAC-seq libraries commonly contain:
- Short fragments, enriched for DNA from accessible regions between nucleosomes.
- Mononucleosome-associated fragments, spanning approximately one nucleosome and adjacent accessible DNA.
- Longer fragments, which can span two or more nucleosomes.
These fragment classes often produce a periodic size distribution. Their relative abundance depends on sample type, preparation, and reaction conditions; they should not be treated as universal, sharply separated categories.
Experimental workflow
Sample preparation
Conventional ATAC-seq begins with cells or isolated nuclei. Preparation must allow the transposase to reach nuclear DNA while preserving chromatin organization. Excessive disruption, damaged cells, or inappropriate detergent conditions can produce high background or altered accessibility profiles.
Nuclei isolation is especially useful for frozen tissues and samples that are difficult to dissociate into intact cells. Nevertheless, sample preservation, storage, and isolation procedures can affect the result, and different materials may require different protocols.
Tagmentation and library construction
Prepared cells or nuclei are exposed to adapter-loaded Tn5 under controlled conditions. The ratio of transposase to input material is important: excessive enzyme activity can overfragment DNA, whereas insufficient activity can yield low-complexity libraries.
Following DNA purification, amplification completes the sequencing adapters and introduces sample indexes. Excessive amplification increases duplication and can distort fragment representation. The final library is assessed for fragment-size distribution, concentration, and suitability for sequencing.
Sequencing and alignment
Sequencing reads are trimmed to remove adapter sequence when necessary and aligned to a reference genome. Processing commonly removes poorly aligned reads, duplicate fragments, and reads from problematic genomic regions.
Reads derived from the mitochondrial genome are usually excluded from nuclear accessibility analysis. Mitochondrial DNA can account for a substantial fraction of some libraries because it is readily exposed during sample preparation and lacks canonical nucleosomal packaging.
When analysis requires precise insertion coordinates, pipelines account for the staggered geometry of Tn5 integration. A common convention shifts positive-strand read starts by bases and negative-strand starts by bases, although coordinate conventions and software implementations must be handled consistently.
Data analysis
Accessibility peaks
A peak is a genomic interval with more ATAC-seq signal than expected from the surrounding background or a statistical model. Peak calling converts aligned fragments or insertion sites into a set of candidate accessible regions.
An accessibility peak does not by itself establish that a region is an enhancer, identify its target gene, or demonstrate a regulatory function. Such interpretations require additional genomic information or experimental evidence.
For comparisons between conditions, fragments are counted within a common set of regions. Statistical models then test for differential accessibility while accounting for sequencing depth, biological variability, and experimental design. Normalization deserves particular care when a perturbation causes broad changes in accessibility: standard methods often assume that most regions do not change substantially.
Regulatory sequence analysis
Accessible regions can be examined for enrichment of DNA sequence motifs recognized by transcription factors. Motif enrichment can suggest regulatory programs associated with a cell type or condition.
However, a motif is not evidence that a particular protein occupies the site. Related transcription factors often recognize similar sequences, and a motif can be present without being bound. Expression measurements and independent protein-binding assays help distinguish among these possibilities.
Footprinting and nucleosome organization
Footprinting attempts to infer protein occupancy from a local depletion of transposase insertions surrounded by more accessible DNA. Such patterns can be informative, but interpretation is complicated by Tn5 sequence preferences, sequencing depth, protein-binding dynamics, and the size of the protein–DNA complex. Reliable footprint analysis requires bias-aware methods and appropriate validation.
Fragment lengths and insertion patterns also support estimates of nucleosome positioning around accessible regions. These estimates are indirect and depend on analytical assumptions; ATAC-seq does not image nucleosomes or resolve every nucleosome in every cell.
Quality assessment
No single metric is sufficient to establish library quality. Common assessments include:
- Transcription start site enrichment: increased signal around annotated sites where transcription begins, relative to flanking background.
- Fraction of reads in peaks: the proportion of usable reads or fragments overlapping called accessible regions. Its value depends partly on the peak set and analysis procedure.
- Library complexity and duplication: the diversity of unique fragments recovered, considered together with sequencing depth.
- Fragment-size distribution: evidence of short-fragment enrichment and, where expected, nucleosome-associated periodicity.
- Mitochondrial read fraction: a measure of sequencing effort diverted from the nuclear genome.
- Replicate agreement: consistency among independent biological samples.
Thresholds are context-dependent. Metrics developed for bulk samples cannot automatically be transferred to individual cells, unusual organisms, or specialized tissues.
Major variants and historical development
The original ATAC-seq method, reported in 2013, demonstrated rapid accessibility profiling from relatively small cell populations. It reduced the preparation burden associated with several earlier genome-wide accessibility assays.
Omni-ATAC, reported in 2017, introduced changes to lysis, washing, and reaction conditions that reduced mitochondrial contamination and background in many sample types. It also broadened practical compatibility with frozen material and tissues. Its performance still depends on the biological material and the quality of preparation.
Single-cell ATAC-seq, demonstrated in 2015, assigns accessible fragments to individual cells or nuclei rather than pooling all signal into a bulk measurement. Implementations use approaches such as microfluidics, droplets, or combinatorial indexing.
Single-cell data are sparse: only a small subset of accessible sites is sampled in each cell. Consequently, analysis usually combines information across many regions and cells to identify populations and regulatory patterns. An absence of reads at a locus in one cell is not sufficient evidence that the locus is inaccessible.
Multimodal methods measure accessibility alongside other molecular features, particularly RNA sequencing, sometimes in the same cell or nucleus. These measurements help connect chromatin state with transcription, although correlations do not by themselves establish causal relationships.
Applications
ATAC-seq is used to characterize cell-type-specific regulatory landscapes, developmental transitions, and responses to experimental perturbations. Single-cell applications are particularly useful for separating accessibility patterns from different cell populations within heterogeneous tissues.
In epigenetics research, the method complements measurements of DNA methylation and histone modifications. Accessibility is one component of chromatin state, not a complete description of it.
Accessible-region maps also assist interpretation of noncoding variants identified by a genome-wide association study. A variant located in a cell-type-specific accessible region may be prioritized for further investigation, but overlap alone does not establish the causal variant, affected gene, or regulatory mechanism.
Integration with genome editing, reporter assays, and targeted perturbations can test the function of candidate regulatory elements. Comparisons with gene-expression data can suggest regulatory relationships, while measurements of chromosomal contacts can provide additional evidence about distant target genes.
Comparison with related methods
DNase-seq maps accessible chromatin using cleavage by DNase I. Both DNase-seq and ATAC-seq provide genome-wide accessibility profiles, but they differ in enzyme biases, sample requirements, and laboratory workflow.
MNase-seq uses micrococcal nuclease digestion and is commonly used to study nucleosome organization. Its signal reflects a different digestion mechanism and is not interchangeable with ATAC-seq accessibility signal.
ChIP-seq enriches DNA associated with a chosen protein or histone modification using an antibody. It addresses molecular occupancy or modification-specific questions, whereas conventional ATAC-seq does not require selecting a particular chromatin protein.
CUT&Tag also uses a transposase, but directs it to an antibody-recognized target. Thus, sharing a tagmentation enzyme does not make CUT&Tag equivalent to ATAC-seq: the former maps a selected molecular feature, while the latter primarily samples accessible DNA.
Limitations and interpretation
ATAC-seq measures the outcome of an enzymatic reaction, not accessibility independently of experimental conditions. Tn5 sequence preferences, chromatin preparation, amplification, read mapping, and sequencing depth can all influence the observed signal.
Bulk profiles average across cells. An apparent increase in accessibility can arise because a region opens within a cell population or because the proportion of a cell type with an already accessible region increases. Single-cell assays help distinguish these explanations but introduce sparsity, doublets, and substantial computational demands.
Highly repetitive regions and sequences that are difficult to map remain challenging. Differences in reference-genome quality and analysis pipelines can also affect comparisons between organisms or studies.
Finally, accessibility alone cannot establish regulatory causation, identify all protein-binding events, or determine which gene a distal element controls. These questions generally require complementary molecular measurements and functional experiments.