Epigenomics FAQ

These answers cover how epigenomics assays and analyses work and how to configure them — ATAC-seq versus ChIP-seq, input controls, replicate counts, FRiP and other quality metrics, narrow versus broad peak calling, bisulfite methylation and DMR calling, RRBS compatibility, differential-method choices, and integrating assay types.

Epigenomics FAQ

What is the difference between ATAC-seq and ChIP-seq?

ATAC-seq measures open chromatin accessibility -- regions where DNA is not tightly wrapped around nucleosomes and is accessible to regulatory proteins. It uses Tn5 transposase to tag accessible DNA. ChIP-seq measures where specific proteins (histone marks or transcription factors) are bound to DNA, using antibody-based immunoprecipitation. ATAC-seq gives a general view of chromatin openness, while ChIP-seq targets specific marks or factors.

Do I need an input control for ChIP-seq?

An input control is strongly recommended but not strictly required. Without input control, the tool relies on a local background model, which may produce more false positive peaks in regions with high mappability or copy number variation. For publication-quality results, always include a matched input control.

How many replicates do I need for differential analysis?

A minimum of 2 biological replicates per condition is required. Three replicates provide substantially better statistical power and are recommended for detecting subtle changes. Technical replicates (re-sequencing the same library) do not substitute for biological replicates and should not be submitted as separate samples.

What is FRiP and why does it matter?

FRiP (Fraction of Reads in Peaks) measures the proportion of sequenced reads that fall within called peaks. It is a key quality metric for ATAC-seq and ChIP-seq experiments. A FRiP above 0.3 for ATAC-seq or 0.1 for ChIP-seq indicates good signal enrichment. Low FRiP values suggest excessive background, poor antibody specificity (ChIP-seq), or insufficient transposase activity (ATAC-seq).

Should I use narrow or broad peak calling?

Use narrow peaks for sharp, well-defined signals: transcription factor binding sites, H3K4me3, H3K27ac, CTCF, and ATAC-seq nucleosome-free regions. Use broad peaks for diffuse histone marks that span large genomic domains: H3K27me3 (Polycomb repression), H3K36me3 (gene body), H3K9me3 (heterochromatin). Using the wrong mode may either fragment a single broad domain into many small peaks or merge distinct narrow peaks into one large region.

How does methylation calling work with bisulfite sequencing?

Bisulfite treatment converts unmethylated cytosines to uracil (read as thymine after PCR), while methylated cytosines remain unchanged. The methylation calling tool compares the observed C/T ratio at each CpG site against the reference genome to calculate the methylation level. A site with 80% C reads and 20% T reads has a methylation level of 0.8. Minimum coverage thresholds (default 5 reads) filter out unreliable sites.

What is a DMR and how is it different from a single CpG?

A DMR (Differentially Methylated Region) is a contiguous genomic region where multiple CpG sites show consistent methylation differences between conditions. While individual CpGs can be noisy, DMRs aggregate signal across multiple sites and are more biologically meaningful. methylation_dmr runs a sliding-window comparison (default window 1,000 bp) over the per-sample methylation files and reports each region's CpG count and methylation difference in the result.

Can I analyze data from reduced representation bisulfite sequencing (RRBS)?

Yes. RRBS data is compatible with the methylation calling and DMR detection tools. RRBS enriches for CpG-dense regions (CpG islands), so genome-wide coverage will be sparse compared to whole-genome bisulfite sequencing (WGBS). The tools handle this correctly by only reporting sites with sufficient coverage.

What differential method should I use?

atacseq_differential defaults to diffbind, with deseq2 as the alternative; the two use different statistical models for calling differentially accessible regions from the peak files you provide. chipseq_differential doesn't expose a method choice. Use the default unless you have a specific reason to compare against the other model.

Can I combine ATAC-seq and ChIP-seq data?

The tools process each assay type independently, but results can be integrated in downstream analysis. Common integrations include overlapping ATAC-seq peaks with ChIP-seq peaks to identify active enhancers (open chromatin + H3K27ac), or correlating accessibility changes with histone mark changes between conditions. Use Ivy to chain analyses and compare results.

Read next