EpiNexus Blog

A free book on epigenomics data analysis

EpiNexus Team — Fri, 23 Oct 2026 12:00:00 GMT

Many researchers who run ChIP-seq, CUT&RUN, CUT&Tag, ATAC-seq or DNA methylation experiments end up with a table of results they can't fully judge. Why did the analysis find 900 changed regions and not 3,000? Is this library good enough to keep? We have written a methods book that answers questions like these: Epigenomics Data Analysis with EpiNexus. It is free to read at epinexus.io/book, and the PDF of edition 0.2 is free to download from Zenodo, with a DOI you can cite.

Why we wrote it

The methods for analysing epigenomic data are well established, but the knowledge of how to use them is scattered. Part of it is in consortium standards such as ENCODE's, part in the code of analysis pipelines, part in tool manuals, and part in the methods sections of many papers. A biologist who wants to know why a step is done a certain way has to search all of them.

The book puts this knowledge in one place. For each step of an analysis it explains what the step does, what experts recommend and why, and how to read the result.

Who it is for

The book is for biologists who analyse their own data, for the staff of core facilities who help them, and for students. You don't need to know how to program. Bioinformaticians may find it useful too, because it gives the source of every published threshold and says where common practice rests on judgement rather than evidence.

The book doesn't cover lab protocols, single-cell methods or chromosome conformation (Hi-C).

What's inside

The book has 21 chapters in six parts, plus four appendices; the PDF has 214 pages.

Foundations: what each assay measures, how to design an experiment you can analyse, and which data and reference files you need.
From reads to peaks: read processing, quality control, peak calling and signal tracks.
Comparing conditions: differential analysis and spike-in normalisation.
From regions to biology: genes and pathways, motifs, enhancers, two histone marks read together, and your regions in public chromatin data.
Case studies: one public dataset per assay, analysed from start to finish.
Reporting: results that others can reproduce, and how to solve common problems.

The appendices hold a glossary, every quality-control threshold with its source, the supported genomes and the file formats.

Every chapter follows the same plan: the question the step answers, how it works, what the field does (with sources), how to do it, how to read a real result, and what can go wrong.

Real data, shown as it came out

Every example uses public data from NCBI GEO, analysed in EpiNexus, and the book shows the results as EpiNexus gave them, including those that didn't turn out as hoped. The case studies include:

H3K27ac ChIP-seq in the mouse heart (GSE155100);
CUT&Tag of H3K27me3 in mouse spermatogonia, with a fly spike-in (GSE221942);
CUT&RUN of the transcription factor MafB in mouse macrophages (GSE287346);
ATAC-seq after removal of SOX2 in mouse embryonic stem cells (GSE209527);
RRBS of neurons lacking the enzymes Dnmt3a and Dnmt3b (GSE167954).

A note on independence

The book's author designed and built EpiNexus, and the book says so in its preface. The methods it teaches are the field's methods, not EpiNexus's own: EpiNexus runs the same tools as published pipelines, such as Bowtie2, MACS3, SEACR, DESeq2, csaw, Bismark, dmrseq and MEME. Published thresholds are kept apart from EpiNexus's own choices, and the book marks each of these choices.

Versions, and how to cite it

Edition 0.2 (October 2026) was written for EpiNexus 1.4.4. The steps that depend on the EpiNexus interface are kept short and name the version they describe, because software changes faster than methods. Each new edition gets its own DOI.

To cite the book:

Osman, A. (2026). Epigenomics Data Analysis with EpiNexus (edition 0.2). https://doi.org/10.5281/zenodo.23472249

You may use its figures and tables in lectures, courses and talks, with credit to the book.

Read the book online, or start with the chapter Quality control you can trust. EpiNexus itself is in an invite-only pilot: ask for access.

Sources

Osman, A. (2026). Epigenomics Data Analysis with EpiNexus, edition 0.2. Zenodo. https://doi.org/10.5281/zenodo.23472249

What's new in EpiNexus since February

EpiNexus Team — Fri, 23 Oct 2026 12:00:00 GMT

In February we wrote about why we built EpiNexus. Since then it has become an invite-only pilot, and almost every step of an analysis has changed. The biggest changes follow the same idea: do each step the way published standards and pipelines do it, and show the researcher where each number comes from. Here are the changes you would notice most, as of EpiNexus 1.5.

Bring in public data without downloading it

You can import data from GEO, SRA or ENA by accession: a series (GSE), a sample (GSM), SRA projects, experiments or runs, or a BioProject or BioSample. EpiNexus lists the samples and fills in each one's target, condition, replicate and controls from the archive's records, for you to confirm. It then copies the reads straight from the archive into your project and checks each file against the archive's MD5 checksum. Runs that ENA holds no FASTQ files for are fetched from SRA and converted. The samples stay editable afterwards.

Quality control graded by published standards

Since EpiNexus 1.2.7, each library is graded against the lines for what it profiled. A narrow histone mark, a broad histone mark, a transcription factor, an ATAC-seq library and a CUT&Tag library each have their own lines, taken from ENCODE and the methods' own papers. Depth is counted in usable fragments, as the standards count it.

There are four grades: Good, Marginal, Below minimum and Fail. Only Fail fails a sample, and only at ENCODE's error level or after a technical failure. Our post Is my ChIP-seq or CUT&Tag data good enough? explains the lines.

Differential analysis as the field does it

DESeq2 in R is the default test, with csaw for broad marks such as H3K27me3.
Sharp histone marks such as H3K27ac are counted over regions from one peak call on all the libraries pooled, as Lun and Smyth recommend. A narrow window around each summit would miss much of a histone mark's signal.
Every pair of conditions is compared, unless you choose which ones. A paired design is available when replicates were processed together.
The whole table of tested regions can be downloaded, not only the significant ones.

From regions to biology

Gene Ontology enrichment follows GREAT's rules: each gene gets a regulatory domain around its start, and regions are tested against the genome, not against a list of nearest genes.
Motifs in changed regions: EpiNexus tests which transcription factor motifs are enriched in gained or lost regions, compared with the other tested regions. Before, it only tested which motifs are common in all peaks, which is a different question.
Enhancer–gene links follow the rules of the activity-by-contact (ABC) model. When the only activity data is H3K27ac, without an accessibility assay, EpiNexus calls the result an ABC-style approximation.
H3K27ac and H3K27me3 read together: when a run holds both marks, each gene is grouped by what both marks did at its promoter.
A genome browser is built in, and each track can also be opened in the UCSC Genome Browser.

The Regulatory Atlas

The largest addition compares your results with public chromatin data of many tissues. It shows what kinds of regulatory element your gained and lost regions fall on, and in which tissues these elements are normally active. The data are:

ENCODE's registry of candidate regulatory elements;
chromatin states of 833 human reference samples (EpiMap) and of mouse fetal tissues;
ENCODE's DNase accessibility of 298 mouse samples.

It works for ChIP-seq, ATAC-seq, CUT&RUN and CUT&Tag runs on human (hg38) and mouse (mm10, mm39) genomes, and starts by itself when a new run finishes.

The Atlas also checks each sample's identity. If you give a sample's tissue on the Samples page, EpiNexus says whether the sample's peaks look like that tissue: Consistent, May not match, Unclear or Not checked. A mismatch can reveal a swapped or mislabelled sample. It is a warning, never a QC failure, and it is done for ATAC-seq and active histone marks such as H3K27ac.

DNA methylation

The library type is detected from the reads. RRBS reads start at the cut site of the enzyme MspI, so EpiNexus can tell RRBS from WGBS before the run starts. RRBS reads are not deduplicated, as the Bismark authors advise.
Differentially methylated regions (DMRs) are called with dmrseq. When none reaches significance, the strongest candidates are shown, labelled as not significant, with the reason.
Context: each DMR is placed on a CpG island, shore, shelf or the open sea, and compared with the CpGs that were tested.
Bisulfite QC sits on the Quality Control page. It grades conversion against ENCODE's 98 %.

Results you can publish and repeat

Methods text is written from what the run actually did, with every tool's version and the references. It comes with a report and a GEO submission package.
Each run records its environment: the software image it ran in, the reference genome and the settings. Run again starts a new run with the same samples and settings on the current software.
Large files, such as signal tracks and aligned reads, download through links, and deleted runs can be restored for 7 days.

Also new

Spike-in normalisation works with any installed genome of another species, and each read is given to one species only.
Genome lists show only the genomes that are installed. Each genome is built from downloads checked against the provider's checksums where the provider publishes them, and human analyses use the no-alt analysis set.

The free book explains each of these steps in depth: Epigenomics Data Analysis with EpiNexus. EpiNexus is in an invite-only pilot: ask for access.

Sources

Lun A.T.L. & Smyth G.K. (2014). De novo detection of differentially bound regions for ChIP-seq data using peaks and windows: controlling error rates correctly. Nucleic Acids Research. https://doi.org/10.1093/nar/gku351
McLean C.Y. et al. (2010). GREAT improves functional interpretation of cis-regulatory regions. Nature Biotechnology. https://doi.org/10.1038/nbt.1630
Fulco C.P. et al. (2019). Activity-by-contact model of enhancer–promoter regulation from thousands of CRISPR perturbations. Nature Genetics. https://doi.org/10.1038/s41588-019-0538-0
Korthauer K. et al. (2019). Detection and accurate false discovery rate control of differentially methylated regions from whole genome bisulfite sequencing. Biostatistics. https://doi.org/10.1093/biostatistics/kxy007
Moore J.E. et al. (2026). An expanded registry of candidate cis-regulatory elements. Nature. https://doi.org/10.1038/s41586-025-09909-9
Boix C.A. et al. (2021). Regulatory genomic circuitry of human disease loci by integrative epigenomics. Nature. https://doi.org/10.1038/s41586-020-03145-z
van der Velde A. et al. (2021). Annotation of chromatin states in 66 complete mouse epigenomes during development. Communications Biology. https://doi.org/10.1038/s42003-021-01756-4

Is my ChIP-seq or CUT&Tag data good enough?

EpiNexus Team — Fri, 23 Oct 2026 12:00:00 GMT

Before you trust a differential result, check each library on three things: is it deep enough, is it complex enough, and is the signal enriched? The first answer is often misread, because "enough reads" depends on what you profiled. ENCODE asks for 45 million usable fragments for a broad mark such as H3K27me3, but 3 million can be enough for a CUT&Tag library. And only a few results mean that a library has actually failed.

1. Depth: count usable fragments, not reads

Standards count usable fragments, not the reads that came off the sequencer. A usable fragment is one DNA molecule whose reads align with good mapping quality, are not duplicates, and don't lie in a blacklisted region. With paired-end reads, the two reads of a fragment count once. A library of 20 million reads with a 70 % mapping rate and 40 % duplicates therefore has far fewer than 10 million usable fragments.

How many you need depends on the assay and the target:

What you profiled	Good	Minimum	Error	Source
ChIP-seq: transcription factor, or a narrow histone mark (H3K27ac, H3K4me3, H3K9ac)	20 M	10 M	under 5 M	ENCODE
ChIP-seq: broad histone mark (H3K27me3, H3K36me3, H3K4me1)	45 M	35 M	under 5 M	ENCODE
ATAC-seq	25 M	20 M	under 15 M	ENCODE
CUT&Tag	3 M	1 M	under 0.2 M mapped reads	Kaya-Okur 2020; Henikoff lab tutorial; Murray 2026
CUT&RUN	3 M	1 M	—	Meers 2019; the minimum is an EpiNexus choice

All values are usable fragments, except the CUT&Tag error line, which a recent benchmark set in mapped reads. H3K9me3 is an exception in ChIP-seq: because it lies mostly in repeats, ENCODE counts all its mapped reads.

CUT&Tag and CUT&RUN need far fewer fragments than ChIP-seq because they have very little background. The developers of CUT&RUN write that they "sequence only 3-5 million fragments, because deeper sequencing provides little if any gain".

Too few fragments make you miss weak changes. They don't make the regions you do find wrong.

2. Complexity: are there enough different molecules?

A library made from too little material holds few distinct DNA molecules. Sequencing it more deeply only reads the same molecules again: the duplicate rate rises, and the number of usable fragments doesn't.

ENCODE measures complexity with three numbers:

NRF: distinct reads divided by all reads.
PBC1 and PBC2: how many positions carry exactly one read, compared with how many carry one or more (PBC1) or exactly two (PBC2).

For ChIP-seq, an NRF of 0.8 or more is good and below 0.5 is below ENCODE's minimum. For ATAC-seq the lines are 0.9 and 0.7.

CUT&Tag and CUT&RUN are different. The enzyme is tethered to the antibody, so it cuts at the same positions in many cells, and identical fragments can be real signal. EpiNexus keeps them in these assays, and shows their duplicate rate for information only.

3. Enrichment: do reads gather where they should?

FRiP, the fraction of reads in peaks, is the most common measure, and the one most often misused.

Transcription factors in ChIP-seq: ENCODE looks more closely at any experiment whose FRiP is below 1 % (Landt et al. 2012).
Histone marks in ChIP-seq: there is no published FRiP value. FRiP depends strongly on how peaks are called, so compare it only between libraries of the same target that were analysed together.
Histone marks in CUT&Tag and CUT&RUN: the nf-core/cutandrun pipeline calls a FRiP above 0.3 (30 %) reasonable.

For ChIP-seq of transcription factors and narrow marks, the strand cross-correlation scores NSC and RSC are a second check. ENCODE repeats replicates below 1.05 and 0.8.

What counts as a failure?

EpiNexus grades each value in one of four ways:

Good: at or better than the good value.
Marginal: between the good value and the minimum.
Below minimum: worse than the minimum.
Fail: worse than the fail value.

Only Fail fails a sample. A fail value exists only where ENCODE sets an error level, such as fewer than 5 million usable fragments for ChIP-seq. A library with no usable fragments, or with peaks but no reads in them, also fails. A library Below minimum is kept, with a warning.

Values that no publication sets, such as EpiNexus's lines for mapping rate, duplicates and peak counts, never fail a sample. This follows ENCODE's own practice: it looks more closely at libraries below its values, but doesn't discard them automatically. A marginal value is a reason to look, not a verdict.

Where to see it in EpiNexus

Since EpiNexus 1.2.7 the Quality Control page grades every library this way:

Dashboard tab: shows each metric, coloured by its grade.
Metrics tab: lists every value, with a Graded as column that says which lines apply to each library (for example "broad histone mark").
Interpretation tab: names each weak metric and suggests what to do.

You decide what happens to failing samples before a run starts. In the Advanced Parameters of the wizard's Parameters step, Samples that fail QC either keeps them and flags the results (the default) or excludes them from the statistics. Excluding leaves out only samples graded Fail.

What to do with a weak library

The metric that is weak usually points to the cause:

What is weak	Likely cause	What to do
Depth only, complexity good	not sequenced deeply enough	sequence the same library again
Complexity	too little material or too many PCR cycles	make a new library; more sequencing won't help
Enrichment	weak antibody or pull-down	repeat the experiment
Mapping rate	contamination, wrong reference genome, adapters, a large spike-in	check the species, genome, trimming and spike-in

With only two replicates per group, think twice before removing one: the group is then left with a single library, and you can no longer see how much replicates differ. Often it's better to keep a marginal library and report it.

The book covers each metric in depth, with a worked example: Quality control you can trust. Every threshold and its source is listed in Appendix B. EpiNexus is in an invite-only pilot: ask for access.

Sources

ENCODE audit code (ENCODE-DCC/encoded, src/encoded/audit/), commit 9bd8ff18 (October 2026); ENCODE data standards, https://www.encodeproject.org/data-standards/.
Landt S.G. et al. (2012). ChIP-seq guidelines and practices of the ENCODE and modENCODE consortia. Genome Research. https://doi.org/10.1101/gr.136184.111
Kaya-Okur H.S. et al. (2020). Efficient low-cost chromatin profiling with CUT&Tag. Nature Protocols. https://doi.org/10.1038/s41596-020-0373-x
Zheng Y. et al. (2020). CUT&Tag data processing and analysis tutorial. protocols.io. https://doi.org/10.17504/protocols.io.bjk2kkye
Murray J.D. et al. (2026). Systematic benchmarking of CUT&Tag improves the reliability and reproducibility of chromatin analysis. Cell Reports Methods. https://doi.org/10.1016/j.crmeth.2026.101499
Meers M.P. et al. (2019). Improved CUT&RUN chromatin profiling tools. eLife. https://doi.org/10.7554/eLife.46314
nf-core/cutandrun, version 3.2.2. https://doi.org/10.5281/zenodo.10606804
Osman, A. (2026). Epigenomics Data Analysis with EpiNexus, edition 0.2, Chapter 6 and Appendix B. https://doi.org/10.5281/zenodo.23472249

Spike-in normalisation: when does it change the answer?

EpiNexus Team — Fri, 23 Oct 2026 12:00:00 GMT

Most ChIP-seq, CUT&Tag and CUT&RUN comparisons don't need a spike-in. Ordinary normalisation works well when most regions stay the same between conditions. A spike-in matters when your treatment may change a mark across the whole genome, for example when a drug blocks the enzyme that makes the mark, or when cells change state. Then ordinary normalisation hides the change, and only a spike-in shows it.

Why ordinary normalisation can't see a global change

A sequencing library doesn't measure how much of a mark a sample holds. Labs make libraries of similar size, and the sequencer reads a similar number of fragments from each. A sample with half as much H3K27me3 still gives about as many reads: they just come from fewer marked molecules.

Normalisation methods such as RLE (DESeq2's default), TMM and library size then make the samples comparable. They all assume that most regions don't change, and they scale each sample so that the typical region looks the same everywhere. When the mark drops everywhere, there is no unchanged typical region, and the scaling removes the loss. A treatment that removes much of a mark can then look as if it changed nothing (Orlando et al. 2014; Chen et al. 2016).

A spike-in is a fixed amount of material from another species, added to every sample at the start: for example, the same number of fly cells in every reaction. It goes through every step with the sample, so its reads show how much those steps scaled each sample. A sample with less of the mark has a larger share of spike-in reads, and is scaled down more.

Three questions before you rely on one

1. Could the treatment change the mark everywhere? Think about the biology: is the enzyme that writes or removes the mark the target of the experiment? Does the cell type change? If the answer is clearly no, ordinary normalisation is the safer choice: a spike-in brings its own sources of error, described below.

2. Was the same amount added to every sample, at the start? A spike-in corrects the data only if it went in equally. As one review puts it, "Spike-in normalization typically makes use of a single scalar to normalize genome-wide data, making the approach particularly vulnerable to errors in implementation" (Patel et al. 2024). A pipetting error shifts every region of that sample.

The E. coli DNA that comes with the CUT&Tag enzyme is a special case. It needs no extra step, but its amount depends on the enzyme batch and on how the enzyme was handled (Kaya-Okur et al. 2019). It works as a spike-in only within one series of samples made with the same enzyme, handled in the same way.

3. Is there enough of it? With few spike-in reads, the scale factor is noisy. The PerCell authors leave out samples whose spike-in share is below 0.5 %, and warn that below about 2 %, "small variations in spike-in content would drastically impact normalization" (Tallan et al. 2025). Too much spike-in, on the other hand, spends sequencing on the other species.

What a global change looks like

A CUT&Tag study of H3K27me3 in mouse spermatogonia gives a clear example (GEO GSE221942; Hu et al. 2024). Each reaction held twice as many fly cells as mouse cells, and there were two replicates of undifferentiated and of differentiating cells. We analysed it with EpiNexus 1.2.3, with the fly genome as the spike-in.

The libraries looked alike. Each had 2.0–2.2 million reads in the regions tested, so ordinary normalisation would have treated the four samples as equal.
The spike-in told them apart. The fly share was 8.3–8.7 % in undifferentiated cells and 11.7 % in differentiating cells. The same number of fly cells went into every reaction, so the differentiating cells gave less H3K27me3 signal. Per spike-in fragment, they had about a quarter fewer reads.
The result was a global loss. With spike-in size factors, 84.5 % of the regions had less H3K27me3 in differentiating cells, and 5,897 regions were called lost against 456 gained.

With ordinary normalisation, the middle of that cloud would have been moved back to zero. The loss would have disappeared, and regions that lost less than most would have looked like gains.

Signs to check in your own results

The MA plot. With spike-in normalisation, a global change shows as a cloud of regions that sits above or below zero. With ordinary normalisation the cloud is centred on zero by design, so a centred cloud says nothing about a global change.
The spike-in share. Compare it between conditions. A consistently larger share in one condition means less of the mark there, if equal amounts went in.
The replicates. After spike-in scaling, replicates should still agree, for example in the sample principal component analysis (PCA). If one replicate lies apart, look at its spike-in share first.
Reads that fit both genomes. Closely related species, such as human and mouse, share sequence. Reads that align to both must not be counted twice (Patel et al. 2024).

How EpiNexus handles spike-ins

In the run wizard, Use spike-in normalization is in the Advanced Parameters of the Parameters step. It is on by default for CUT&Tag and CUT&RUN, with E. coli as the spike-in genome. For other spike-ins, choose the species you added from the installed genomes. If the GEO records of imported samples mention a spike-in, the wizard suggests it.

During the run, EpiNexus:

gives each read to the genome it aligns to better, and drops reads that fit both equally well. This rule comes from sequencing human tumours grown in mice (Ahdesmäki et al. 2016);
for ChIP-seq where every sample has its own input, corrects each factor by the spike-in ratio in the input (Fursova et al. 2019; Patel et al. 2024); otherwise uses the spike-in counts alone;
notes a spike-in share below 0.5 % or 2 % (from PerCell; not for E. coli, whose share is usually small), above 20 %, or with more than 5 % of spike-in reads fitting both genomes. The last two are EpiNexus rules;
calibrates only samples with at least 1,000 spike-in fragments, an EpiNexus rule. If any sample in a comparison has fewer, that comparison falls back to RLE, and the run says so.

The Quality Control page lists each sample's spike-in fragments, share and calibration, and the Differential page names the normalisation that was actually used. To see what the spike-in changes in your own data, start a second run with the same samples and the spike-in switched off, and compare the two MA plots.

The book covers spike-ins in more depth, with the full example: Spike-in normalisation. EpiNexus is in an invite-only pilot: ask for access.

Sources

Orlando D.A. et al. (2014). Quantitative ChIP-Seq normalization reveals global modulation of the epigenome. Cell Reports. https://doi.org/10.1016/j.celrep.2014.10.018
Chen K. et al. (2016). The overlooked fact: fundamental need for spike-in control for virtually all genome-wide analyses. Molecular and Cellular Biology. https://doi.org/10.1128/MCB.00970-14
Patel L.A. et al. (2024). The Wild West of spike-in normalization. Nature Biotechnology. https://doi.org/10.1038/s41587-024-02377-y
Kaya-Okur H.S. et al. (2019). CUT&Tag for efficient epigenomic profiling of small samples and single cells. Nature Communications. https://doi.org/10.1038/s41467-019-09982-5
Tallan A. et al. (2025). Highly quantitative measurement of differential protein-genome binding with PerCell chromatin sequencing. Cell Reports Methods. https://doi.org/10.1016/j.crmeth.2025.101052
Hu M. et al. (2024). PRC1 directs PRC2-H3K27me3 deposition to shield adult spermatogonial stem cells from differentiation. Nucleic Acids Research. https://doi.org/10.1093/nar/gkad1203
Ahdesmäki M.J. et al. (2016). Disambiguate: an open-source application for disambiguating two species in next generation sequencing data from grafted samples. F1000Research. https://doi.org/10.12688/f1000research.10082.2
Fursova N.A. et al. (2019). Synergy between variant PRC1 complexes defines Polycomb-mediated gene repression. Molecular Cell. https://doi.org/10.1016/j.molcel.2019.03.024

DNA Methylation Data Analysis with EpiNexus

EpiNexus Team — Fri, 20 Feb 2026 12:00:00 GMT

Updated 3 October 2026 to match the current EpiNexus app.

DNA methylation — the covalent addition of a methyl group to the 5-carbon of cytosine, predominantly at CpG dinucleotides — is one of the most extensively studied epigenetic modifications in biology. It plays a central role in gene regulation, genomic imprinting, X-chromosome inactivation, transposon silencing, and genome stability. Aberrant methylation patterns are hallmarks of cancer, neurodevelopmental disorders, and aging.

Modern technologies now allow genome-wide methylation profiling at single-base resolution. But for most wet lab researchers, turning those raw sequencing files into biological insight remains a serious computational challenge. In this post, we walk through what methylation analysis involves, why it's hard, and how EpiNexus makes it accessible.

How methylation is measured

There are several experimental approaches to profiling DNA methylation, each with trade-offs in resolution, coverage, and cost:

Whole-genome bisulfite sequencing (WGBS) treats genomic DNA with sodium bisulfite, which converts unmethylated cytosines to uracils while leaving methylated cytosines intact. After sequencing and alignment, you get the methylation status of virtually every CpG in the genome — roughly 28 million sites in humans. WGBS is the gold standard for unbiased, comprehensive methylome profiling, but requires deep sequencing (30x or more per sample).

Reduced representation bisulfite sequencing (RRBS) uses restriction enzyme digestion (typically MspI) to enrich for CpG-dense regions before bisulfite treatment. This captures around 1–3 million CpGs concentrated at promoters and CpG islands, at roughly a tenth of the sequencing cost of WGBS — making it practical for large cohort studies.

Enzymatic methyl-seq (EM-seq) is a newer alternative that replaces the harsh bisulfite conversion with enzymatic reactions (TET2 oxidation followed by APOBEC deamination). The result is the same base-resolution methylation readout, but with less DNA degradation, better coverage uniformity, and improved performance on limited or degraded input DNA such as FFPE samples or cell-free DNA.

Methylation arrays (Illumina 450K, EPIC) measure methylation at hundreds of thousands of predefined CpG sites using probe intensities, reporting beta values (proportion methylated) or M-values (logit-transformed). While not sequencing-based, they remain widely used in clinical and epidemiological studies.

All sequencing-based approaches produce FASTQ files that look like standard data — but require specialized alignment and quantification to handle the bisulfite-converted bases.

The traditional analysis pipeline

A typical methylation analysis chains together several command-line tools, each with its own installation headaches, parameter choices, and output formats:

1. Quality control and trimming. FastQC and Trim Galore assess read quality and remove adapters. Methylation data requires additional attention to end-of-read bias — bisulfite conversion introduces systematic artifacts at read termini that can skew methylation estimates if not properly trimmed.

2. Bisulfite-aware alignment. Tools like Bismark or bwa-meth align reads to an in-silico converted reference genome. The aligner must consider both C-to-T and G-to-A converted versions of the genome simultaneously. This makes alignment significantly slower and more memory-intensive than standard DNA-seq mapping. Post-alignment QC includes checking mapping rates, coverage uniformity, and M-bias plots to detect position-specific artifacts.

3. Methylation extraction. Bismark's methylation extractor or MethylDackel calculates the methylation percentage at each cytosine from the ratio of unconverted (C) to converted (T) reads. Strand-specific quantification is important for proper downstream analysis.

4. Filtering and summarization. CpGs with insufficient coverage (typically fewer than 10 reads) are removed, along with sites overlapping known SNPs. Methylation values can be aggregated over genomic windows, CpG islands, promoters, or other features of interest.

5. Exploratory analysis. PCA, hierarchical clustering, and global methylation distributions help detect batch effects, sample swaps, and broad biological patterns before diving into differential analysis.

6. Differential methylation. Statistical tools like DSS, methylKit, or dmrseq identify differentially methylated CpGs (DMCs) and differentially methylated regions (DMRs) between conditions. Region-level methods often provide higher statistical power and more biologically interpretable results by smoothing or aggregating signals from neighboring CpGs.

7. Annotation and interpretation. DMRs are mapped to genomic features — promoters, gene bodies, enhancers, CpG islands, shores, and shelves — and tested for enrichment in pathways, gene ontology terms, and regulatory element databases.

For a researcher who just received sequencing data from the core facility, assembling and validating this pipeline can take weeks.

What makes methylation data challenging

Beyond the usual bioinformatics hurdles, methylation data has specific complexities that make analysis particularly demanding:

Reduced sequence complexity. Bisulfite conversion turns most unmethylated C's into T's, effectively reducing the four-letter DNA alphabet. This makes read alignment slower, more ambiguous, and prone to mapping artifacts — especially in repetitive regions.

Coverage is uneven and expensive. WGBS requires deep sequencing for reliable single-CpG resolution, and even at 30x average coverage, some regions will have gaps. RRBS is cheaper but only covers a subset of CpGs. Researchers must make informed decisions about which sites to include and how to handle low-coverage data.

Subtle biological signals. Unlike histone modifications where a peak is present or absent, methylation changes are often quantitative shifts — from 80% to 60% methylated, for example. Detecting these requires statistical models that properly account for the binomial nature of count data and biological variability across replicates. Methods must handle overdispersion and apply careful multiple-testing correction across millions of sites.

Genomic context matters enormously. A CpG losing methylation in a promoter has very different biological implications than one in a gene body, intergenic region, or transposable element. CpG islands, shores (2 kb flanking regions), and shelves (the next 2 kb beyond) each carry distinct functional significance. Analysis that ignores context misses much of the biology.

Cell-type composition confounds bulk data. Bulk tissue samples mix cells with different methylation profiles. An apparent methylation change between conditions might actually reflect a shift in cell-type proportions rather than true epigenetic reprogramming. Deconvolution or reference-based approaches may be needed to disentangle these effects.

How EpiNexus handles methylation

EpiNexus wraps the methylation pipeline into the same workflow used for ChIP-seq and other epigenomic assays:

Upload your data

Drag and drop your FASTQ.gz files from WGBS or RRBS experiments, or import published samples from GEO, SRA or ENA by accession. If you have already aligned your reads with Bismark, you can start from its coverage (.cov) files instead. Select your reference genome, put each sample in a group — treatment vs. control, tumor vs. normal, or any comparison — and choose the reference group.

Before the run starts, EpiNexus checks whether your libraries are RRBS or WGBS from the reads themselves: RRBS reads begin at an MspI site (CGG or TGG), WGBS reads begin anywhere. The Review step shows the result, and you can choose the library type yourself instead.

Automated pipeline

Once you click run, EpiNexus executes the full pipeline:

Trimming — adapter and quality trimming with Trim Galore; for RRBS, its --rrbs mode removes the filled-in bases at MspI-site ends
Bisulfite-aware alignment — Bismark (Bowtie2) against C-to-T and G-to-A converted copies of the genome
Deduplication — PCR duplicates are removed for WGBS; RRBS is never deduplicated, as Bismark advises, because every read of a fragment starts at the same MspI site
Methylation calling — per-CpG methylation with Bismark's methylation extractor, with the two strands of each CpG merged; M-bias plots show any read-end bias, and for paired-end WGBS the first 2 bp of read 2 are ignored, as the Bismark documentation recommends
Differential methylation — differentially methylated regions (DMRs) with dmrseq, which models biological variation across replicates and controls the false discovery rate at the region level; by default it tests CpGs covered by at least 4 reads in every sample and calls regions at FDR ≤ 0.05, and you can change both values for a run. Without replicates, EpiNexus falls back to a per-CpG Fisher test and labels those results exploratory.
Annotation — every DMR is linked to its nearest gene, and placed in its CpG-island context (island, shore, shelf or open sea; UCSC islands, for 13 vertebrate genomes), compared with all tested CpGs

Explore your results

The Methylation page shows:

An overview — how many CpGs were tested, mean methylation in each group, and the global difference between them. When no region reaches significance, the page says so, shows how many candidate regions dmrseq found against how many the sample swaps produced, and lists the strongest candidates, labelled as not significant.

A volcano plot — methylation difference on the x-axis, statistical significance on the y-axis, hypermethylated and hypomethylated regions in different colours.

Chromosome distribution, CpG-island context and a DMR table — where the DMRs fall, with their size, number of CpGs, methylation difference, FDR and nearest gene; the table downloads as a tab-separated file.

Samples — a PCA and a correlation of the samples, their methylation and coverage distributions, and how many CpGs remain at each coverage cut-off.

Bisulfite QC, on the Quality Control page — for each sample: reads, mapping rate, duplicates removed, mean CpG coverage, methylation in CpG, CHG and CHH context, an estimate of bisulfite conversion (graded against ENCODE's 98% standard), and M-bias plots. A MultiQC report collects the tools' own reports.

The report and the Export page include the methylation figures, ready for a paper.

Putting methylation in context

Methylation changes are most meaningful when viewed alongside other layers of epigenomic regulation:

Methylation and histone marks — a promoter that loses methylation and gains H3K4me3 suggests transcriptional activation.
Methylation and chromatin accessibility — demethylation at a region that becomes accessible in ATAC-seq points to active regulatory remodeling.
Pathways and motifs — the genes near DMRs, and the transcription-factor motifs within them, suggest which processes and regulators are affected.

You can analyse ChIP-seq and ATAC-seq data from the same system in EpiNexus and compare the genes near your DMRs with those results. A joint methylation–chromatin analysis, and GO or motif enrichment of DMRs, are not part of EpiNexus yet.

Choosing the right approach for your experiment

The right methylation method depends on your question and practical constraints:

WGBS is best for comprehensive, unbiased discovery — when you need to survey the entire methylome, including enhancers, gene bodies, and intergenic regions. Budget 400–800 million reads per sample for mammalian genomes.

RRBS is ideal for focused studies on promoters and CpG islands with many samples. At 30–50 million reads per sample, it's an order of magnitude cheaper than WGBS and well-suited for clinical cohorts.

EM-seq gives WGBS-level coverage with better data quality, especially valuable for limited or degraded input DNA — FFPE tissue, liquid biopsies, rare cell populations.

EpiNexus processes WGBS and RRBS with the same upload-and-click workflow, detecting which one you have from the reads and adapting trimming and deduplication to it. EM-seq isn't supported as its own library type yet.

Getting started

DNA methylation analysis doesn't need to be a months-long computational project. If you have FASTQ files from a WGBS or RRBS experiment, EpiNexus can take you from raw reads to differentially methylated regions and their nearest genes, with bisulfite QC, all from your browser.

Methylation analysis is available in the EpiNexus pilot at app.epinexus.io. Accounts are by invitation — if you'd like one, get in touch; we'd love to hear about your project.

Further reading: For a comprehensive review of methylation analysis methods, see Bock (2012) in Nature Reviews Genetics. For hands-on pipeline tutorials, the Galaxy Training Network and the Computational Genomics with R book provide excellent walkthroughs.

Why We Built EpiNexus

EpiNexus Team — Thu, 19 Feb 2026 12:00:00 GMT

Updated 3 October 2026 to match the current EpiNexus app.

Epigenomic experiments generate rich, complex data — ChIP-seq peaks, histone modification maps, super-enhancer landscapes. But for most wet lab researchers, turning raw FASTQ files into publication-ready figures means either learning command-line bioinformatics or waiting weeks for a collaborator.

We built EpiNexus to close that gap.

The problem

A typical ChIP-seq analysis involves a dozen tools: FastQC, Trimmomatic, Bowtie2, SAMtools, MACS2, DiffBind, ROSE, HOMER, deepTools — each with its own installation headaches, parameter choices, and output formats. Even experienced bioinformaticians spend days stitching these together.

For a biologist who just wants to know which enhancers changed between treatment and control, this is an unreasonable barrier.

Our approach

EpiNexus wraps the entire pipeline — from raw FASTQ trimming and read mapping through peak calling, quality control, genomic annotation, differential analysis, and super-enhancer detection — into a single web interface. You upload your files (or import published data from GEO, SRA or ENA), choose your genome and assay type, and click run.

No terminal. No scripts. No conda environments.

What you get

Every analysis produces interactive, publication-ready outputs:

Volcano plots showing gained and lost peaks with gene labels
Signal heatmaps across all your replicates
Super-enhancer hockey stick plots ranked by signal
Genomic annotation charts breaking down where your peaks land
Downloadable tables ready for supplementary materials
The Regulatory Atlas: your changed regions compared with public chromatin data of many tissues, and a check that each sample looks like its tissue

Figures are interactive — hover for details — and the main ones export as SVG or PNG for your paper. Every run also records its tool versions and commands and writes a methods section you can adapt.

Who it's for

EpiNexus is built for wet lab researchers who generate epigenomic data but don't want to become bioinformaticians. If you can upload a file and click a button, you can run a full analysis.

We currently support ChIP-seq, CUT&Tag, CUT&RUN, ATAC-seq, and DNA methylation (WGBS and RRBS).

What's next

EpiNexus is now open as an invite-only pilot at app.epinexus.io. If you'd like an account, get in touch — we'd love to hear about your project.