Answers to common questions about EpiNexus, the analyses it runs, and the assays it supports.
EpiNexus is a web-based platform for analysing epigenomic data. You upload your sequencing files (FASTQ, or BAM with peak files) or import them from GEO, SRA or ENA, choose a genome, and click run. EpiNexus handles alignment, peak calling, quality control, peak annotation, differential analysis, GO and motif enrichment, multi-mark integration, super-enhancer detection, and DNA-methylation analysis, and compares your results with public chromatin data of many tissues (the Regulatory Atlas) — all without writing any code.
No. EpiNexus is designed for biologists who generate ChIP-seq or related data but don't want to learn command-line tools. Everything is done through a point-and-click web interface.
Sign up at app.epinexus.io with your institution’s email address and start a free trial: one analysis of up to 8 libraries within 30 days. Free webmail and disposable addresses can’t be used. For full access, write to info@epinexus.io. EpiNexus is a commercial platform — the same address is for licensing and demos.
EpiNexus runs as a web application at app.epinexus.io. Open it in any modern browser (Chrome, Firefox, Safari, or Edge), sign in (or create an account), and create a project to get started. If you forget your password, use “Forgot password?” on the sign-in page.
Yes. Each user has their own account and projects. A project belongs to the person who created it, so your data and results stay separate from other users.
EpiNexus starts from raw reads or from aligned reads:
FASTQ files — Raw sequencing reads straight from the sequencer. If you have these, EpiNexus handles the entire workflow: alignment, peak calling, and downstream analysis.
BAM files with peak files — Aligned sequencing reads (already mapped to a genome), together with the peaks called from them (.narrowPeak or .broadPeak). EpiNexus calls peaks only from FASTQ files, so a BAM-based analysis needs your peak files too.
Bismark coverage files — For DNA methylation, if your bisulfite reads are already aligned with Bismark.
Public data — Samples published in GEO, SRA or ENA can be imported by accession; the reads go straight from the archive into your project.
No problem! Choose the Full Pipeline when creating your run. EpiNexus will align the reads to the genome, call peaks, and run the full analysis automatically. You don't need to do any preprocessing yourself.
Full Pipeline for FASTQ files: everything from reads to results. Preprocessing Only for FASTQ files when you want to check data quality first: trimming, alignment, peaks, signal tracks and QC, without differential or downstream analyses. Analysis Only for BAM files uploaded with their peak files. The Review step tells you if your files don't fit the pipeline you chose.
FASTQ files contain the raw sequence reads exactly as they came off the sequencer — each read is a short DNA fragment (~75–150 bp) with quality scores. BAM files are the same reads but mapped to a reference genome, so each read now has a genomic coordinate. Peak files are the result of peak calling — they list the specific genomic regions where your mark is enriched, condensing millions of reads into thousands of peaks.
FASTQ: 5–30 GB per sample (paired-end generates two files). BAM: 1–10 GB per sample. Peak files: 0.5–20 MB (very small). Each account has 100 GB of storage. A stable internet connection helps; if an upload is interrupted, it resumes from where it stopped.
An input or IgG control improves peak calling accuracy, but it is not strictly required. If you have one, register it as a control on the Samples page and EpiNexus uses it as the background for the samples of the same condition and replicate; one control can also serve several samples, such as a shared IgG (controls are never called for peaks themselves). If you don't, EpiNexus still calls peaks against a background model.
For basic QC and annotation, one sample is enough. For differential analysis (comparing conditions), you need at least two replicates per condition — ideally three or more. More replicates give you better statistical power to detect real changes.
EpiNexus supports five assays:
ChIP-seq — The standard method for profiling histone modifications and transcription factor binding. This is what EpiNexus was built for.
CUT&Tag — A newer, low-input alternative to ChIP-seq. Works well for histone marks, especially in samples with limited cell numbers.
CUT&RUN — Another low-input chromatin profiling method. For CUT&Tag and CUT&RUN, EpiNexus follows the CUT&Tag data-processing tutorial: SEACR peak calling by default (GoPeaks optional), duplicates kept in target samples, and E. coli spike-in calibration.
ATAC-seq — Measures open chromatin accessibility rather than histone marks. EpiNexus adds ATAC-specific processing and QC (Tn5 shift, mitochondrial reads removed, fragment sizes, TSS enrichment) and runs annotation, differential accessibility, GO and motif enrichment. Super-enhancer calling is not run on ATAC-seq.
DNA methylation — Bisulfite sequencing, WGBS or RRBS (detected from the reads). EpiNexus aligns with Bismark, finds differentially methylated regions (DMRs) with dmrseq, and reports each DMR's nearest gene and CpG-island context, with bisulfite QC.
Any histone mark that produces peaks. Commonly used marks include:
H3K27ac — Active enhancers and promoters. Ideal for super-enhancer analysis.
H3K4me1 — Primed and active enhancers.
H3K4me3 — Active promoters.
H3K27me3 — Polycomb-repressed regions.
H3K9me3 — Heterochromatin / constitutive silencing.
H3K36me3 — Gene bodies of actively transcribed genes.
EpiNexus picks the peak shape from the mark: narrow peaks for sharp marks such as H3K27ac and H3K4me3, broad domains for H3K27me3, H3K9me3, H3K36me3 and H3K4me1. You can also analyse transcription factor ChIP-seq (e.g. p53, CTCF, ERα), which is called with narrow peaks.
Yes. A project can hold samples for several marks, and each mark is analysed in its own right. Put H3K27ac and H3K27me3 samples in one run, compared between the same conditions, and the Integration page reads the two marks together at each gene's promoter — for example “H3K27ac gained, H3K27me3 lost” — with a gene table you can download.
Not currently. EpiNexus is specialised for epigenomic assays — histone modifications, chromatin accessibility, and DNA methylation. RNA-seq and WGS require different analysis pipelines.
Runs that start from FASTQ files take several hours, mostly for alignment, depending on the number of samples and sequencing depth. Runs that start from BAM files are shorter. You can close your browser and come back later — the analysis runs on the server, and the run page shows each step's progress.
If you start from FASTQ files (Full Pipeline), EpiNexus first trims, aligns and filters the reads and calls peaks. If you start from BAM files (Analysis Only), it uses your BAM and peak files. Then it runs the downstream steps:
1. Quality Control — Usable fragments, mapping, duplication, library complexity, FRiP, fragment sizes and TSS enrichment, graded against published standards for each kind of sample.
2. Replicate reproducibility — Which peaks replicate (IDR or overlap).
3. Annotation — Labels each peak by genomic feature and finds the nearest gene.
4. Peak–Gene Links — Connects regulatory regions to the genes they likely control (ABC model), for ATAC-seq, DNase-seq or H3K27ac samples.
5. Differential Analysis — Compares conditions to find regions that gain or lose signal (DESeq2, or csaw for broad marks).
6. GO Enrichment — Gene Ontology terms near gained, lost, or all peaks.
7. Multi-Mark Integration — H3K27ac and H3K27me3 changes read together at each gene's promoter.
8. Super-Enhancers — ROSE, for H3K27ac, MED1 or BRD4 samples.
9. TF & Motif Enrichment — JASPAR motifs enriched in your peaks, and in the regions that gained or lost signal.
10. Signal Profiles — Heatmaps and profiles around peaks (runs from FASTQ files).
11. MultiQC Report — The tools' QC reports in one place.
After the run: Regulatory Atlas — Your regions and samples compared with public chromatin data (human hg38, mouse mm10 and mm39).
All steps run automatically; those that don't apply to your data are skipped, and the run page says why. DNA-methylation runs follow their own path (bisulfite alignment, DMR calling, MultiQC). See the Pipelines page for details.
Yes. QC, annotation, peak–gene links, GO and motif enrichment, and super-enhancers (for H3K27ac, MED1 or BRD4 samples) all work with a single condition. Only differential analysis needs two or more conditions to compare.
A differential peak is one whose signal changes significantly between two conditions (e.g. treated vs. untreated). "Gained" peaks have more signal in the treatment, "lost" peaks have less. This helps you identify the genomic regions affected by your experimental condition.
Super-enhancers are clusters of nearby enhancers with exceptionally high levels of an active mark (typically H3K27ac) or of co-activators such as MED1 and BRD4. They drive the expression of cell-identity genes and are frequently disrupted in disease. EpiNexus identifies them in H3K27ac, MED1 or BRD4 data using the ROSE algorithm, which ranks enhancers by signal strength and finds the inflection point separating super-enhancers from typical enhancers.
Fraction of Reads in Peaks — the percentage of your sequencing reads that fall within called peaks. What counts as good depends on the assay and the target, so EpiNexus grades it only where a published value exists: transcription factors in ChIP-seq (good from 5%, below the minimum under 1%), ATAC-seq (good from 30%, below the minimum under 20%, ENCODE) and histone marks in CUT&Tag and CUT&RUN (good from 30%, nf-core/cutandrun). For histone marks in ChIP-seq, ENCODE sets no FRiP value, so EpiNexus shows it without a grade. Low FRiP may indicate low enrichment, too much background, or a failed experiment.
Each metric is Good, Marginal (between the good value and the minimum), Below minimum or Fail. Below minimum is a warning: the sample is kept. A sample fails only at ENCODE's error level — for example fewer than 5 million usable fragments in ChIP-seq — or when it has no usable data. The values depend on the kind of sample: a broad mark such as H3K27me3 needs 45 million usable fragments, a narrow mark such as H3K27ac 20 million. Values that no publication sets are EpiNexus's own choices; they never fail a sample.
False Discovery Rate — the expected share of false positives among the results called significant. With an FDR cut-off of 0.1, about 10% of the peaks called differential are expected to be false positives. EpiNexus calls a peak differential when its FDR is below 0.1 and its fold change is more than √2 (|log2 fold change| > 0.5); for DNA-methylation regions the cut-off is FDR ≤ 0.05 by default, and you can change it for a run.
Yes. Tables download as tab-separated files, which open in Excel or Google Sheets; the full differential results download as a CSV; the main figures download as PNG or SVG. Under Output, Reports builds a report with figures and a methods section, and Export downloads the run's peaks and tables as a zip, a GEO submission package, and the files to reproduce the run. Signal tracks, alignments and methylation calls are too large for the zip; Export lists them as direct download links, valid for 24 hours. Every file a run produced is also listed on its Results tab.
EpiNexus is configured for 17 genomes covering the most widely used model organisms:
Human — hg38 (GRCh38), hg19 (GRCh37)
Mouse — mm39 (GRCm39), mm10 (GRCm38)
Rat — rn7 (mRatBN7.2), rn6 (Rnor_6.0)
Zebrafish — danRer11 (GRCz11)
Fruit fly — dm6 (BDGP6)
C. elegans — ce11 (WBcel235)
Yeast — sacCer3 (R64)
Chicken — galGal6 (GRCg6a)
Macaque — rheMac10 (Mmul_10)
Pig — susScr11 (Sscrofa11.1)
Cow — bosTau9 (ARS-UCD1.2)
Dog — canFam6 (UU_Cfam_GSD_1.0)
Frog — xenLae2 (v10.1)
Arabidopsis — tair10 (TAIR10)
When you create a project, the app lists the genomes installed on the service. See the Pipelines page for the reference sources of each.
Check with whoever aligned your data (usually your core facility). For recent human data, hg38 (GRCh38) is the standard. For mouse, mm39 (GRCm39) is the latest. If you're unsure, look at the BAM file header or the aligner output — it will mention the genome reference used.
Write to us at info@epinexus.io. We can install any of the 17 configured genomes that isn't offered yet, and we'll look at adding other organisms that have a reference genome and a gene annotation.
Open the run page: the steps table shows which step failed and why, and the Logs tab has the details. The most common issues are:
Wrong genome selected — Make sure the genome in your project matches the genome your FASTQ/BAM files were aligned to.
Corrupted or partial upload — Try re-uploading the file. “Transfer verified” next to a file confirms a complete upload.
Very low quality data — If very few peaks are detected, the downstream steps may have nothing to work with. Check the QC results for low FRiP scores.
Missing replicates for differential — Differential analysis requires at least two samples per condition.
If you're stuck, email us at info@epinexus.io with the run details and we'll help troubleshoot.
If alignment and peak calling fail, the run stops, because every later step needs their output. If a later step fails, the other steps still run and finish: for example, if differential analysis fails, annotation and super-enhancer detection still complete normally. The run page shows which step failed and why; fix the issue and re-run.
Yes, any finished run (completed, failed or cancelled): click Delete run on the run, or Delete on the project’s Runs page. It disappears at once and can be restored from Recently deleted for 7 days; then its result files are removed for good, and backup copies expire within a further 30 days. A run that is still going must be cancelled first.
Yes. Open All Projects, find the project you want to remove, and click its trash icon. This permanently deletes the project, all uploaded files, and all runs and results; backup copies expire within 30 days. You will be asked to confirm before anything is deleted. A project with a run still going can’t be deleted until that run is cancelled or finished.
Yes. Within the same project, you can create as many runs as you like — for example with different samples, comparisons, normalisation or peak caller. Your uploaded files are reused automatically, and each analysis page lets you switch between runs. Run again on a finished run repeats it with the same samples and settings, using the current software.
Every run records the EpiNexus version, the software it ran in, every setting and every command. The report and the methods text name the version that analysed the data. A run keeps its results when EpiNexus is updated; Run again uses the current version, so its results can differ where a method changed.
FASTQ and BAM files can be several gigabytes. Use a wired connection if possible, and avoid uploading during peak network hours. If an upload is interrupted, add the same file again and it resumes from where it stopped. For published data, importing from GEO, SRA or ENA avoids the upload entirely.
A peak is a region of the genome where your histone mark or factor is enriched above background. Think of it as a "hotspot" of epigenetic activity. Each peak has a location (chromosome, start, end) and a signal strength. A typical ChIP-seq experiment yields thousands to tens of thousands of peaks.
Narrow peaks are sharp, well-defined regions (typically a few hundred base pairs). They are characteristic of transcription factors and marks like H3K4me3 and H3K27ac. Broad peaks are wider domains (tens of kilobases) produced by marks like H3K27me3, H3K9me3, and H3K36me3. EpiNexus handles both types automatically.
TSS = Transcription Start Site, the position where a gene begins to be transcribed. The distance from a peak to the nearest TSS tells you whether the peak is at a promoter (within ~3 kb) or a distal enhancer (often 10–1000 kb away).
When one run has differential results for both the active mark H3K27ac and the repressive mark H3K27me3, between the same two conditions, EpiNexus looks at each gene's promoter (3 kb either side of its start). For each mark it notes whether the promoter gained, lost, did both in different regions, or did not change, using that mark's own differential results. The gene then gets one group:
H3K27ac gained, H3K27me3 lost and H3K27ac lost, H3K27me3 gained — the two marks changed in opposite directions (shown as “Opposite changes”).
H3K27ac gained only, H3K27ac lost only, H3K27me3 gained only, H3K27me3 lost only — one mark changed and the other showed no significant change (with two replicates, a real change can be missed).
Both gained, Both lost, Mixed changes, No change.
Changes in regions within 100 kb of a gene, but not on its promoter, are shown beside the groups as “nearby”. A flow chart shows how the promoter calls of the two marks combine, and the gene table can be downloaded.
ROSE (Rank Ordering of Super-Enhancers) is the standard method for identifying super-enhancers. It works by: (1) leaving out peaks at promoters (within 2.5 kb of a TSS), (2) stitching together nearby enhancer peaks within 12.5 kb, (3) ranking all stitched enhancers by their signal, and (4) finding the inflection point in the ranking curve. Regions above the inflection point are classified as super-enhancers.
The Activity-by-Contact (ABC) model predicts which gene each enhancer regulates from two factors: the enhancer's activity (its signal in your data) and how often it contacts the gene's promoter in 3D. EpiNexus estimates contact from the genomic distance (a power law, as the ABC authors do when no Hi-C data is available) and links enhancers to genes up to 5 Mb away. Activity comes from ATAC-seq, DNase-seq or H3K27ac; with H3K27ac alone the links are labelled an ABC-style approximation. This is usually better than simply taking the nearest gene, especially for distal enhancers.
After a run, EpiNexus compares your changed regions and each sample's peaks with public chromatin data: ENCODE's registry of candidate regulatory elements, the chromatin states of hundreds of reference samples from many tissues, and genome-wide chromatin states. It shows, for example, that the regions you lost are active enhancers in heart. It covers human (hg38) and mouse (mm10, mm39). The data come in versioned releases from sources whose licences allow this use; the app's Data sources page lists every source and what to cite.
On the Samples page you can give each sample its tissue and a reference tissue. After the run, EpiNexus compares each sample's peaks with the reference tissues of the Regulatory Atlas and says whether the sample looks like its reference 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. It is done for ATAC-seq and active histone marks such as H3K27ac; marks such as H3K27me3 are not checked.