Direct Epigenetics and Methylation Analysis with Long-Read Sequencing — No Bisulfite, Single-Molecule Resolution
A cancer researcher preparing precious tumor biopsy DNA for methylation analysis faces a dilemma. The gold-standard method — bisulfite conversion followed by sequencing — will destroy 84-96% of her input DNA through chemical degradation. If her tumor sample carries a differentially methylated CpG island within a LINE-1 element, she may never detect it. If one parental allele is hypermethylated and the other is not, bisulfite-treated short reads cannot phase them — the haplotype-specific methylation signal collapses into a population average.
Bisulfite conversion, the workhorse of DNA methylation analysis for two decades, has three fundamental limitations that long-read sequencing eliminates: DNA fragmentation requiring high input amounts, inability to distinguish 5mC from 5hmC (opposite biological functions), and loss of haplotype-specific methylation information. Long-read sequencing solves all three by reading native, unconverted DNA molecules in their entirety.
Why researchers choose our epigenetics and methylation analysis services
- No bisulfite — native DNA preserved
- Haplotype-resolved methylation across kilobase-scale reads
- Single-molecule modification detection and stoichiometry quantification
- Chromatin accessibility + DNA methylation from one experiment
Why Direct Epigenetic Detection Matters — Beyond Bisulfite Conversion
Bisulfite conversion cannot distinguish 5-methylcytosine (5mC) from 5-hydroxymethylcytosine (5hmC) — both read as cytosine after conversion — yet these marks have opposite biological functions: 5hmC is an active demethylation intermediate enriched in neurons and embryonic stem cells, while 5mC is typically associated with gene silencing. Second, bisulfite treatment fragments DNA, requiring high input amounts that exclude low-yield samples such as cell-free DNA, FFPE biopsies, and single-cell preparations. Third, short-read bisulfite sequencing loses haplotype information: a chromosome region where the maternal allele is fully methylated and the paternal allele is fully unmethylated appears as 50% methylation — indistinguishable from a region where both alleles are 50% methylated. Long-read direct detection avoids all three limitations by reading native DNA without chemical conversion.
How Long-Read Sequencing Detects Epigenetic Marks Without Chemical Treatment
Long-read sequencing platforms detect epigenetic modifications through fundamentally different physical mechanisms, neither of which requires chemical conversion of the DNA template.
PacBio HiFi sequencing detects 5mC through polymerase kinetics. During SMRT (Single Molecule, Real-Time) sequencing, a DNA polymerase incorporates fluorescently labeled nucleotides while a zero-mode waveguide captures the light pulse from each incorporation event. The inter-pulse duration (IPD) changes measurably when the polymerase encounters a modified base. Because HiFi reads are generated by circular consensus sequencing (multiple passes over the same molecule), kinetic signals are averaged across passes, producing high-confidence methylation calls at single-base, single-molecule resolution. Critically, this kinetic signature differs between 5mC and 5hmC, enabling their distinction in one run. The same sequencing run simultaneously provides genomic sequence and methylation status of every CpG site, covering >90% of CpGs genome-wide.
Oxford Nanopore Technologies (ONT) detects modifications through direct measurement of ionic current changes. As a single DNA or RNA molecule passes through a nanopore protein, it partially blocks the ionic current. Each nucleotide modification produces a characteristic disruption in the current signal. Deep learning models (DeepMod2, Nanopolish, Megalodon, Dorado) trained on these signal patterns can call modifications at single-base resolution. ONT detects 5mC, 5hmC, and 6mA in DNA; m6A, pseudouridine, inosine, and other modifications in RNA. The key advantage: ONT reads RNA directly —no reverse transcription, no PCR —preserving the native modification landscape.
The practical implication: one sequencing experiment now replaces what previously required two or three separate workflows —genetic variant calling from the sequence data, methylation calling from the modification signal, and haplotype-resolved methylation retaining parental origin of each methylated CpG.
Epigenetics and Methylation Analysis Services at CD Genomics
CD Genomics provides four specialized epigenetics and methylation analysis services within our LongSeq division. Each service page includes detailed workflows, sample requirements, bioinformatics deliverables, and project consultation information.
Direct DNA Modification Detection
Long-Read Sequencing of DNA Methylation —Genome-wide 5mC and 5hmC detection at single-base, single-molecule resolution using PacBio HiFi (polymerase kinetics) or ONT (nanopore signal). Covers whole-genome methylation profiling, CpG island methylation analysis, differentially methylated region (DMR) identification, and haplotype-resolved methylation phasing. Unlike bisulfite, direct detection preserves native DNA, distinguishes 5mC from 5hmC, and covers >90% of genomic CpG sites —including repetitive regions where bisulfite-treated reads cannot map uniquely. Deliverables include methylation tracks (bedMethyl), DMR lists, and phased methylation visualization.
Direct RNA Modification Detection
Long-Read Sequencing of RNA Methylation —Transcriptome-wide RNA modification detection and quantification using ONT direct RNA sequencing. Detects m6A, pseudouridine, inosine, and additional RNA modifications without antibody enrichment, bisulfite conversion, or reverse transcription. Key advantage over antibody-based methods (MeRIP-seq, m6A-CLIP): single-nucleotide resolution and modification stoichiometry quantification —we report what fraction of molecules carry the modification at each site, not just presence/absence.
Multi-Omics Chromatin Analysis
Fiber-seq Service —Simultaneous chromatin accessibility mapping and DNA methylation detection on single DNA molecules. Uses m6A-methyltransferase to label accessible chromatin regions with m6A marks; ONT sequencing reads both synthetic m6A (chromatin accessibility) and endogenous 5mC (DNA methylation) from the same molecule. Replaces three separate experiments: ATAC-seq, WGBS, and WGS. For allele-specific chromatin regulation and epigenetic reprogramming studies, Fiber-seq reveals mechanistic relationships that separate assays cannot.
Custom Methylation Detection Pipeline
ONT-gDNA-RRMS —A customized ONT-based workflow for genomic DNA methylation detection with flexible bioinformatics configuration. Designed for researchers who need optimized methylation calling for specific genomic contexts —repetitive elements, imprinted loci, or allele-specific methylation. We configure methylation calling tools (DeepMod2, Megalodon, Nanopolish, Dorado) based on your chemistry version (R9.4.1 or R10.4.1), organism, and research question.
For epigenetics services beyond LongSeq, visit CD Genomics' Epigenetics, Epitranscriptome & Chromatin Analysis Platform. For technology information, see PacBio SMRT Sequencing Technology and Oxford Nanopore Sequencing Technology.
DNA Methylation Analysis —Haplotype-Resolved, Genome-Wide, No Bisulfite
DNA methylation is the most intensively studied epigenetic mark, yet most researchers have only seen it through the lens of bisulfite conversion —a lens that distorts what it measures. Our long-read DNA methylation service addresses specific analytical gaps that bisulfite-based methods cannot close.
CpG Coverage Beyond Array Limits. The Illumina EPIC array interrogates approximately 850,000 CpG sites —less than 3% of the ~28 million CpGs in the human genome. Whole-genome bisulfite sequencing (WGBS) improves coverage but loses reads: after bisulfite reduces sequence complexity (all unmethylated Cs become Ts), 10-20% of reads fail to map uniquely —disproportionately from repeat-rich regions. PacBio HiFi and ONT long reads span repetitive regions with flanking unique sequence, mapping unambiguously and recovering methylation information genome-wide.
Haplotype-Resolved Methylation. This is the most important capability that long reads unlock. In imprinted genes, one parental allele is methylated and silenced while the other is unmethylated and expressed. In cancer, allele-specific methylation is a hallmark of loss of imprinting. Bisulfite sequencing cannot phase CpG methylation across haplotypes because reads are too short to link neighboring SNPs to CpGs. Long reads, spanning tens of kilobases, naturally link SNPs to CpG methylation on the same molecule, producing phased calls that distinguish maternal from paternal alleles.
5mC vs 5hmC Distinction. 5-hydroxymethylcytosine is a stable epigenetic mark enriched in Purkinje neurons (~40% of modified cytosines), embryonic stem cells, and brain tissue, with distinct regulatory functions. PacBio HiFi kinetic signatures distinguish 5mC and 5hmC in a single sequencing run —no need for parallel oxidative bisulfite libraries.
RNA Modification Analysis —Single-Molecule Detection Without Antibody Enrichment
The epitranscriptome —the complete set of RNA modifications across all transcripts —has emerged as a critical regulatory layer in gene expression. More than 170 distinct RNA modifications are known, with m6A the most abundant internal modification in eukaryotic mRNA and pseudouridine the most abundant overall.
The Antibody Problem. MeRIP-seq, the dominant method for transcriptome-wide m6A detection, uses an anti-m6A antibody to enrich modified RNA fragments. This approach has three limitations: resolution is poor (100-200 nt fragments), antibody specificity varies (cross-reactivity with m6Am), and —most critically —it cannot measure modification stoichiometry. An enriched peak could represent 90% modification or 5% modification —biologically different scenarios indistinguishable by antibody-based methods.
ONT Direct RNA Sequencing solves these problems by reading native, full-length RNA molecules directly through a nanopore. In a single experiment, you obtain: (1) full-length transcript sequence enabling isoform identification, (2) location and identity of RNA modifications via tools such as Nanopolish, Tombo, or EpiNano, and (3) modification stoichiometry —the fraction of reads showing modification at each position. No reverse transcription, no antibody enrichment, no amplification bias.
For researchers validating epitranscriptomic targets from MeRIP-seq, ONT direct RNA sequencing provides the orthogonal confirmation that reviewers increasingly expect —single-nucleotide resolution, stoichiometry quantification, and isoform-level modification mapping from one experiment.
Fiber-seq —Simultaneous Chromatin Accessibility and DNA Methylation on Single Molecules
Fiber-seq addresses a fundamental limitation in epigenomics: chromatin accessibility and DNA methylation have historically been measured in separate experiments, on separate sample aliquots, using different biochemical principles. ATAC-seq tells you which regions are nucleosome-depleted but nothing about methylation. WGBS tells you which CpGs are methylated but nothing about chromatin state. Integrating these datasets requires statistical inference, not direct measurement.
Fiber-seq changes this by combining both measurements on the same DNA molecule. The workflow uses m6A-methyltransferase (EcoGII or Hia5) to deposit m6A marks specifically on accessible DNA —regions not protected by nucleosomes or bound proteins. ONT sequencing then detects both the synthetic m6A marks (chromatin accessibility) and endogenous 5mC marks (DNA methylation) from the same reads.
A single Fiber-seq molecule reveals:
- Nucleosome positions (gaps in m6A labeling)
- Transcription factor footprints (~10-30 bp protected regions within accessible DNA)
- DNA methylation status at every CpG
- Genetic variants (SNPs, indels, SVs) along the same molecule
For allele-specific regulation studies, Fiber-seq is uniquely powerful. Traditional approaches require phasing SNPs to assign ATAC-seq reads to haplotypes, then separately phasing methylation calls —a statistical exercise with compounding error. Fiber-seq reads naturally link accessibility, methylation, and genotype on each molecule.
Platform Selection for Epigenetics — PacBio HiFi vs Oxford Nanopore
Choosing between PacBio HiFi and Oxford Nanopore for epigenetics depends on your specific modification types, required accuracy, and experimental design.
| Feature | PacBio HiFi | Oxford Nanopore (ONT) |
| Methylation Detection | Polymerase kinetics (IPD ratio) | Ionic current signal + deep learning |
| DNA Modifications Detected | 5mC, 5hmC, 6mA | 5mC, 5hmC, 6mA |
| RNA Modifications | Not applicable | m6A, pseudouridine, inosine, m5C, and others |
| Detection Resolution | Single-base, single-molecule | Single-base, single-molecule |
| Accuracy (5mC Calling) | >95% at CpG sites (HiFi consensus) | >90% (model-dependent, improves with coverage) |
| 5mC vs 5hmC Distinction | Yes — distinct kinetic signatures | Emerging (model-dependent) |
| Read Length | 15-25 kb (HiFi) | 50 kb – 1+ Mb |
| Haplotype Phasing | Excellent — high accuracy for trio-binning | Good — ultra-long reads span large blocks |
| Modification Stoichiometry | Yes — per-molecule kinetic signal | Yes — per-read signal classification |
| Fiber-seq Support | Not supported | Supported — m6A-MTase labeling |
| Direct RNA Sequencing | Not available | Available — preserves native modifications |
| Recommended For | DNA methylation reference maps, 5hmC studies, phased methylation discovery | RNA modification analysis, Fiber-seq multi-omics, extreme repetitive regions |
Decision Guide:
- Need genome-wide DNA methylation with 5hmC distinction → PacBio HiFi (kinetic signatures distinguish these marks)
- Need RNA modification detection (m6A, pseudouridine, inosine) → ONT direct RNA sequencing
- Need chromatin accessibility + methylation simultaneously → ONT with Fiber-seq (unique to this platform)
- Need custom methylation detection pipeline → ONT-gDNA-RRMS
Our project consultation team helps you select the right platform — or combination — based on your research question, sample type, and required deliverables.
Sample Requirements for Epigenetics and Methylation Analysis
Epigenetic analysis places specific demands on sample quality because modification detection requires intact, native nucleic acids. Chemical degradation, freeze-thaw cycles, and enzymatic activity can alter or erase the very marks you are trying to measure.
| Sample Type | Quantity | Quality Requirement | Special Handling |
| Genomic DNA (gDNA) | 2-5 μg | HMW (>40 kb), OD 260/280 1.8-2.0 | Avoid vortexing; wide-bore tips; no freeze-thaw |
| Total RNA | 1-5 μg | RIN ≥ 8.0; DNase-treated | Flash-freeze immediately; RNAlater optional |
| Cultured Cells | 1-5 × 106 cells | Viability >90% | Pellet, flash-freeze, ship on dry ice |
| Fresh Tissue | 50-100 mg | Flash-frozen within 30 min of collection | Avoid RNAlater for methylation samples |
| FFPE Tissue | 5-10 scrolls (10 μm) | Documented fixation time | HMW DNA recovery varies; contact for feasibility |
| Blood | 2-5 mL | EDTA tube (not heparin); non-coagulated | Ship on cold packs within 48 h |
| Plant Tissue | 2-5 g (young leaf) | Flash-frozen; polysaccharide-rich species need CTAB extraction | High-polyphenol samples require PVP treatment |
Critical considerations for long-read methylation projects:
- HMW DNA is essential. Short or degraded DNA reduces read length and methylation phasing capability. We recommend HMW extraction using phenol-chloroform or magnetic bead protocols optimized for long fragments.
- No bisulfite treatment. Do not bisulfite-convert your samples before submission — our direct detection methods work on native DNA/RNA.
- Avoid RNase contamination for RNA modification projects. Even trace RNase activity degrades long RNA molecules needed for isoform-level modification analysis.
For full sample preparation guidelines, visit our Sample Submission Guideline page.
Bioinformatics for Epigenetic Data — From Raw Signal to Biological Insight
Raw nanopore signal data or PacBio kinetic data requires specialized bioinformatics pipelines to translate electrical current traces or polymerase kinetics into biologically interpretable methylation calls. CD Genomics provides end-to-end analysis tailored to your chosen platform and research question.
DNA Methylation Calling:
- PacBio HiFi: We use pb-CpG-tools and Primrose for 5mC detection from kinetic data, with ccs generating HiFi reads. 5hmC detection uses KineticTools with 5hmC-specific models. Output: bedMethyl files, per-site modification probability, strand-specific methylation calls.
- ONT gDNA: We deploy platform-appropriate callers based on flow cell chemistry. For R10.4.1 data, Dorado with the 5mC_5hmC model provides state-of-the-art accuracy. For R9.4.1 data, DeepMod2 or Megalodon with guppy basecalling. All pipelines output modification frequency at single-base resolution.
RNA Modification Analysis (ONT direct RNA-seq): Modification detection uses Nanopolish (eventalign + signal-level analysis), Tombo, or EpiNano depending on the modification type. For m6A detection, we use tools trained on IVT vs native RNA comparisons. Modification stoichiometry is calculated as the fraction of reads showing modification signal at each position. Key deliverables: modification position, modification type, modification fraction (stoichiometry), transcript isoform association, and comparative differential modification analysis.
Fiber-seq Multi-Omics Analysis: Data processing uses Fibertools: m6A-MTase label detection (chromatin accessibility), 5mC detection (DNA methylation), nucleosome position inference from labeling gaps, and transcription factor footprint detection. Output: single-molecule chromatin tracks (Fiber-seq BAM), nucleosome occupancy profiles, TF footprint coordinates, and combined accessibility-methylation visualization.
Standard deliverables for all epigenetics projects include: raw signal data (fast5/pod5), basecalled reads (FASTQ), aligned reads (BAM), modification calls (bedMethyl or equivalent), differential modification analysis (if multiple conditions), and a comprehensive bioinformatics report with QC metrics. Visit our Long-Read Sequencing Data Analysis Services hub for complete analysis offerings.
FAQs
Q: Why not just use bisulfite conversion for DNA methylation?
Bisulfite conversion works, but it has three hard limitations that affect biological conclusions. First, bisulfite destroys 84-96% of input DNA through chemical degradation — problematic for low-input samples. Second, it cannot distinguish 5mC from 5hmC; both appear as protected cytosines after conversion, yet they have opposite biological functions (5hmC is enriched in active chromatin, 5mC in silenced regions). Third, short bisulfite-treated reads cannot be phased — you lose haplotype-specific methylation, which is critical for imprinting studies, allele-specific expression, and cancer epigenetics. Long-read direct detection avoids all three limitations by reading native DNA without chemical treatment, covering >90% of CpGs genome-wide while simultaneously capturing genetic variants and methylation on the same molecule.
Q: Can you distinguish 5mC from 5hmC?
Yes. PacBio HiFi kinetic signatures differ measurably between 5mC and 5hmC because the hydroxymethyl group produces a distinct perturbation in polymerase translocation kinetics compared to the methyl group. ONT signal-based tools are also increasingly capable of this distinction, though accuracy depends on the basecalling model and coverage depth. If distinguishing 5mC from 5hmC is critical to your project, we recommend PacBio HiFi — this is one of the strongest differentiators between the two platforms.
Q: Do you support non-model organisms without reference genomes?
Yes. Our methylation calling pipelines on ONT (DeepMod2, Megalodon) and PacBio (pb-CpG-tools) do not require a reference genome for modification detection — they analyze raw signal/kinetic data against the sequenced reads themselves. For reference-free methylation analysis, we first perform de novo genome assembly using our Animal/Plant De Novo Sequencing or Microbial Genomics services, then map modification calls to the new assembly. For RNA modification analysis in non-model species, we use reference-free isoform discovery tools (FLAIR, Bambu) followed by modification calling on identified transcripts.
Q: Which platform — PacBio HiFi or ONT — should I choose?
If your primary goal is high-accuracy DNA methylation mapping with 5mC/5hmC distinction, choose PacBio HiFi. If you need RNA modification detection or chromatin accessibility + methylation multi-omics (Fiber-seq), choose ONT. If you need maximum read length for spanning centromeric repeats or large structural variants while capturing methylation, choose ONT ultra-long reads. Many projects combine both platforms — our team helps you design the optimal strategy during project consultation.
Q: What RNA modifications can you detect with direct RNA sequencing?
ONT direct RNA sequencing detects m6A (the most abundant internal mRNA modification), pseudouridine, inosine, 5-methylcytosine (m5C), and other modifications that produce detectable changes in nanopore ionic current. The specific modifications detectable depend on the basecalling model and training data — we use the latest Dorado models and supplement with modification-specific tools (Nanopolish, Tombo, EpiNano). Unlike antibody-based methods, we quantify modification stoichiometry: the fraction of transcripts carrying the modification at each site. For modifications not included in standard models, custom training on synthetic modified/unmodified RNA pairs is available.
Q: How does Fiber-seq compare to separate ATAC-seq and WGBS?
Fiber-seq measures chromatin accessibility and DNA methylation on the same single DNA molecule — ATAC-seq and WGBS measure them on separate aliquots of different cells. This single-molecule co-measurement reveals regulatory relationships that separate assays miss: a nucleosome variably positioned on the maternal allele is invisible to bulk ATAC-seq (you see average occupancy, not allele-specific differences), and methylation differences linked to allele-specific accessibility require statistical phasing with associated error. Fiber-seq directly links accessibility, methylation, and genotype on individual reads. Furthermore, Fiber-seq requires only one library preparation and one sequencing run, reducing cost, input requirements, and batch effects compared to three separate assays.
Q: What are your sample requirements?
For DNA methylation projects, we require 2-5 μg of high-molecular-weight DNA (OD 260/280 1.8-2.0, >40 kb fragment size). For RNA modification projects, we need 1-5 μg of total RNA (RIN ≥ 8.0). Both should be flash-frozen and shipped on dry ice. Avoid bisulfite treatment, vortexing, and freeze-thaw cycles. See our Sample Submission Guideline page for detailed protocols.
Q: How do I start a project?
Contact our epigenetics consultation team through the inquiry form. To provide an accurate project proposal, we need: (1) your research question and the epigenetic marks you are interested in, (2) species and estimated genome size, (3) sample type, quantity, and storage conditions, (4) number of samples and experimental groups, (5) whether you need DNA methylation, RNA modifications, chromatin accessibility, or a combination, and (6) any timeline constraints. We respond with a customized project proposal including platform recommendation, sequencing strategy, bioinformatics plan, timeline, and quotation — typically within 1-2 business days.
Demo
1. CpG Coverage Comparison — Bisulfite Array vs Long-Read Direct Detection: Bar chart comparing genomic CpG coverage: Illumina EPIC array (~3%, ~850K sites), WGBS (~60%, mapping loss in repeats), PacBio HiFi methylation (~95%, whole genome including repeats).
2. Haplotype-Resolved Methylation at an Imprinted Locus: Genome browser-style track showing phased methylation reads at the H19/IGF2 imprinted locus. Maternal allele reads show dense CpG methylation at the ICR (imprinting control region); paternal allele reads show unmethylated CpGs — clearly visible on long reads spanning the entire locus.
3. RNA Modification Stoichiometry Quantification: Scatter plot showing m6A stoichiometry (Y-axis, 0-100%) across individual transcripts (X-axis, ranked by expression). Dotted line at 50% indicates the population-average modification level; the spread of individual transcripts above and below this line reveals the single-molecule stoichiometry information that antibody-based methods cannot provide.
References
- Ahsan MU, Gourle H, Vadiakas G, et al. Computational analysis of DNA methylation from long-read sequencing. Nature Reviews Genetics, 2025.
- Gershman A, Sauria MEG, Lhoumaud P, et al. Profiling the epigenome using long-read sequencing. Nature Genetics, 57, 10-21 (2025).
- Ahsan MU, Gourle H, Vadiakas G, et al. DeepMod2: A deep learning framework for methylation detection from nanopore sequencing. Nature Communications, 15, 1548 (2024).
*For research use only. Not for use in diagnostic procedures.*