ATAC-seq peak calling is the algorithmic identification of statistically significant enrichment of sequencing reads over background, corresponding to nucleosome-free regions of open chromatin. The process analyzes transposase-accessible chromatin sequencing data to map genome-wide chromatin accessibility, revealing where transcription factors and other regulatory proteins can bind to DNA.
Glossary
ATAC-seq Peak Calling

What is ATAC-seq Peak Calling?
ATAC-seq peak calling is the computational process of identifying genomic regions of open, accessible chromatin from sequencing data, pinpointing active regulatory elements like promoters and enhancers.
Peak callers like MACS2 and Genrich model the fragment size distribution characteristic of open chromatin, distinguishing true signal from Tn5 transposase insertion bias and background noise. The resulting peaks represent candidate cis-regulatory elements, including active promoters, enhancers, and insulators, forming the foundation for downstream differential accessibility analysis.
Key Features of a Peak Caller
An ATAC-seq peak caller is a computational algorithm designed to identify regions of open chromatin from sequencing data. These tools must distinguish true biological signal from background noise, account for the Tn5 transposase binding bias, and provide statistically rigorous enrichment scores.
Fragment Size Distribution Analysis
A critical pre-processing step that leverages the biological signature of nucleosome-free and nucleosome-bound regions. True open chromatin signal is enriched for sub-nucleosomal fragments (<100 bp), while nucleosome-protected regions generate longer fragments (~200 bp). Peak callers analyze this bimodal distribution to filter out background noise and focus on high-confidence regulatory elements.
Tn5 Bias Correction
The Tn5 transposase exhibits an inherent sequence insertion bias, preferentially cutting at specific nucleotide motifs. Failure to correct for this bias leads to false-positive peaks at intrinsically favorable sequences. Advanced peak callers incorporate bias models trained on genomic DNA controls or use k-mer-based correction algorithms to normalize the signal and ensure peaks reflect true chromatin accessibility rather than enzymatic preference.
Statistical Enrichment Testing
Peak callers employ rigorous statistical frameworks to distinguish signal from background. Common approaches include:
- Poisson distribution models for count-based enrichment
- Negative binomial models to account for overdispersion in biological replicates
- Hidden Markov Models for segmenting the genome into accessible and inaccessible states
- Permutation-based FDR control to establish genome-wide significance thresholds
Replicate Concordance and IDR
The Irreproducible Discovery Rate framework is the gold standard for assessing peak reproducibility across biological replicates. It evaluates the consistency of peak rankings between replicate pairs, retaining only peaks that appear in both datasets with high confidence. This approach directly addresses the false discovery problem inherent in high-throughput genomic assays and is required by ENCODE consortium standards.
Peak Shape and Footprinting
Beyond simple peak detection, sophisticated callers analyze the local read density profile around binding sites. Transcription factor footprints appear as narrow regions of depleted signal flanked by accessible chromatin, revealing the precise location of protein-DNA binding. This sub-peak analysis transforms a binary accessibility map into a high-resolution view of regulatory complex occupancy.
Blacklist and Mappability Filtering
Reference genome artifacts, such as high-mappability regions and segmental duplications, generate spurious signal that mimics true peaks. Production-grade peak callers integrate empirically derived blacklists that flag problematic genomic intervals. These regions are excluded from downstream analysis to prevent false-positive regulatory annotations that would mislead functional interpretation.
Frequently Asked Questions
Clear, technically precise answers to the most common questions about identifying open chromatin regions from ATAC-seq data.
ATAC-seq peak calling is the computational process of identifying genomic regions of open, accessible chromatin from sequencing data generated by the Assay for Transposase-Accessible Chromatin using sequencing (ATAC-seq). The process works by detecting local enrichment of sequencing read fragments, which occur because the hyperactive Tn5 transposase preferentially inserts sequencing adapters into nucleosome-free DNA. A peak caller algorithm scans the genome in sliding windows, comparing the density of mapped read start sites against a background model of random distribution. Regions where fragment accumulation statistically exceeds the expected background are identified as peaks, representing putative regulatory elements such as promoters, enhancers, and insulators. The core statistical challenge is distinguishing true biological signal from technical artifacts like mitochondrial DNA contamination and PCR duplication.
Comparison of Peak Calling Algorithms
A feature-level comparison of widely used algorithms for identifying open chromatin regions from ATAC-seq data, highlighting algorithmic approach, fragment handling, and output characteristics.
| Feature | MACS2 | HMMRATAC | Genrich |
|---|---|---|---|
Algorithmic Approach | Poisson distribution with local bias correction | Hidden Markov Model with semi-supervised learning | Statistical model with multiple testing correction |
Nucleosome Fragment Handling | Shifts reads by Tn5 insertion offset | Models fragment length distribution directly | Removes PCR duplicates and filters by MAPQ |
Paired-End Support | |||
Single-End Support | |||
Broad Peak Detection | |||
ATAC-seq Specific Optimization | |||
Peak Significance Metric | q-value (Benjamini-Hochberg) | Posterior probability | p-value with multiple testing correction |
Runtime (50M reads) | < 5 min | 30-60 min | < 3 min |
Enabling Efficiency, Speed & Accuracy
Intelligent Analysis, Decision & Execution
We build AI systems for teams that need search across company data, workflow automation across tools, or AI features inside products and internal software.
Talk to Us
Search across company data
Give teams answers from docs, tickets, runbooks, and product data with sources and permissions.
Useful when people spend too long searching or get different answers from different systems.

Automate internal workflows
Use AI to route work, draft outputs, trigger actions, and keep approvals and logs in place.
Useful when repetitive work moves across multiple tools and teams.

Add AI to products and internal tools
Build assistants, guided actions, or decision support into the software your team or customers already use.
Useful when AI needs to be part of the product, not a separate tool.
Related Terms
Explore the core computational concepts and deep learning architectures that underpin ATAC-seq peak calling and the broader field of regulatory genomics.
Chromatin State Annotation
The systematic categorization of genomic segments into functional states like promoters, enhancers, or insulators based on combinatorial histone modification and protein binding patterns.
- Integrates data from ChIP-seq, DNase-seq, and ATAC-seq
- Uses Hidden Markov Models or ChromHMM to segment the genome
- Defines the regulatory grammar that peak callers aim to identify
Sequence-to-Epigenome Modeling
A deep learning paradigm where a model predicts genome-wide epigenomic tracks, such as chromatin accessibility or histone marks, solely from raw DNA sequence input.
- Bypasses the need for wet-lab assays in silico
- Core architectures include Enformer and Basenji2
- Learns the cis-regulatory code that governs peak formation
Enformer Network
A transformer-based deep learning architecture from DeepMind that predicts gene expression and epigenomic tracks from DNA sequence with long-range attention up to 200 kilobases.
- Captures distal enhancer-promoter interactions
- Significantly improves prediction accuracy over previous models
- Directly informs peak calling by modeling the determinants of accessibility
In-Silico Mutagenesis
A computational perturbation technique that systematically introduces virtual mutations into a DNA sequence to quantify their predicted impact on an epigenomic model's output.
- Identifies causal regulatory variants within called peaks
- Scores every possible single-nucleotide change for its effect on accessibility
- Essential for prioritizing non-coding variants in disease studies
Epigenomic Latent Space
The compressed, high-dimensional vector representation learned by an autoencoder or foundation model that captures the underlying structure of complex epigenomic data.
- Encodes the regulatory syntax of the genome
- Can be used to cluster similar peaks or cell types
- Enables cross-cell-type generalization for peak calling in unassayed contexts
Integrated Gradients
A model interpretability method that attributes the prediction of a genomic neural network to its input sequence by accumulating gradients along a path from a baseline to the actual input.
- Produces nucleotide-resolution importance scores
- Reveals the sequence motifs driving a peak call
- Satisfies the completeness axiom, ensuring all attribution is accounted for

About the author
Prasad Kumkar
CEO & MD, Inference Systems
Prasad Kumkar is the CEO & MD of Inference Systems and writes about AI systems architecture, LLM infrastructure, model serving, evaluation, and production deployment. Over 5+ years, he has worked across computer vision models, L5 autonomous vehicle systems, and LLM research, with a focus on taking complex AI ideas into real-world engineering systems.
His work and writing cover AI systems, large language models, AI agents, multimodal systems, autonomous systems, inference optimization, RAG, evaluation, and production AI engineering.
Partnered with leading AI, data, and software stack.
How We Work
Custom AI workflows for your Business
One-fit-all AI don't work for modern businesses. At Inferensys, we aim to understand your business & custom requirements; which we use to define most efficient agentic workflows, the data, and the tools for your business.
01
Review the use case
We understand the task, the users, and where AI can actually help.
Read more02
Pick the right approach
We define what needs search, automation, or product integration.
Read more03
Build the first useful version
We implement the part that proves the value first.
Read more04
Improve from there
We add the checks and visibility needed to keep it useful.
Read moreThe first call is a practical review of your use case and the right next step.
Talk to Us