Nanopore Direct RNA Sequencing — Full-Length Native RNA Sequencing for Direct Transcriptome Analysis & RNA Modification Detection

Base modification detection (m⁶A, m⁵C, Ψ, inosine), poly(A) tail profiling, and full-length native transcript isoform analysis by Oxford Nanopore dRNA-seq on PromethION with RNA004 chemistry — no reverse transcription, no PCR amplification, no amplification bias

CD Genomics provides Nanopore Direct RNA Sequencing (dRNA-seq) on PromethION with RNA004 chemistry — the only commercially available RNA sequencing method that sequences native RNA molecules directly without reverse transcription or PCR amplification. Every read simultaneously delivers full-length transcript sequence, single-nucleotide RNA modification detection (m⁶A, m⁵C, Ψ, inosine), and poly(A) tail length measurement from the same unamplified native RNA molecule.

Every RNA sequencing method that relies on reverse transcription and PCR amplification introduces a fundamental trade-off: you gain throughput, but you lose the native RNA molecule. The cDNA copy is not the original — it carries the biases of reverse transcriptase (premature termination at structured regions, GC-dependent processivity, template-switching artifacts), the distortions of PCR duplication (amplification bias, loss of modification information), and the blind spots of fragmentation (loss of connectivity between distal transcript features). For transcriptome-wide RNA modification analysis, poly(A) tail profiling, and full-length native transcript characterization, these are not merely technical limitations — they are systematic information losses that no amount of computational correction can recover.

CD Genomics offers Nanopore Direct RNA Sequencing (dRNA-seq) — the only commercially available RNA sequencing method that sequences native RNA molecules directly. Using the Oxford Nanopore platform with RNA004 chemistry on PromethION and GridION instruments, our dRNA-seq service delivers full-length native transcript sequences, single-nucleotide RNA modification detection (m⁶A, m⁵C, pseudouridine Ψ, inosine, and 2′-O-methylation), poly(A) tail length measurements, and transcript isoform identification from the same unamplified RNA molecules. No reverse transcription bias. No PCR distortion. No fragmentation. The RNA molecule itself is the sequencing template.

Why Nanopore Direct RNA Sequencing — Service Highlights

• Only commercially available RNA-seq method that sequences native RNA — Direct RNA sequencing reads the original RNA molecule without reverse transcription or PCR amplification, preserving every native base modification and RNA structural feature as it exists in the biological sample. This is the only platform that provides simultaneous sequence, modification, and poly(A) tail information from the same unamplified molecule
• Multi-modality data from a single sequencing experiment — Each dRNA-seq read simultaneously delivers five data types: (1) full-length transcript sequence from 5′ cap to 3′ poly(A) tail, (2) single-nucleotide modification probabilities for m⁶A, m⁵C, Ψ, inosine, and 2′-O-methylation, (3) poly(A) tail length at single-nucleotide resolution, (4) splice isoform structure with exon connectivity, and (5) transcript-level expression quantification — all from one library preparation and sequencing run
• Amplification-free detection of base modifications — Because the native RNA molecule is never reverse-transcribed or PCR-amplified, every base modification present in the original sample is preserved and detectable. Modified bases produce characteristic ionic current signatures as they pass through the nanopore, enabling direct identification of m⁶A, m⁵C, Ψ, and other modifications at single-nucleotide resolution without antibodies, chemical treatment, or enrichment steps
• Full-length native transcript sequencing from 5′ to poly(A) tail — Each dRNA-seq read represents a complete native RNA molecule, spanning from the 5′ cap through the entire transcript body to the 3′ poly(A) tail. There is no fragmentation step, no size selection bias, and no computational transcript reconstruction — each read is a direct observation of a full-length native transcript with its complete splice architecture, modification landscape, and poly(A) tail status
• Poly(A) tail measurement at single-nucleotide resolution — The poly(A) tail, a critical regulator of mRNA stability, translation efficiency, and decay, is captured and measured directly from the raw ionic current signal as the native RNA passes through the nanopore. Poly(A) tail length distributions are obtained at single-nucleotide resolution for every sequenced transcript, enabling direct correlation between tail length, isoform usage, and modification status on individual RNA molecules
• RNA004 chemistry on PromethION for maximum yield — We deploy the latest ONT RNA004 kit chemistry on PromethION and GridION instruments, achieving significantly improved sequencing yields for direct RNA applications compared to earlier R9.4.1 chemistries. The RNA004 system employs an optimized motor protein and adapter system specifically designed for native RNA, delivering more reads per flow cell and supporting a broader range of RNA input amounts

Why Direct RNA Sequencing — and Why It Changes How We See the Transcriptome

The eukaryotic transcriptome is not a simple collection of linear sequences. Every mature mRNA carries a complex layer of regulatory information encoded in its base modifications, poly(A) tail length, splice isoform choice, and 5′/3′ untranslated region architecture — all of which are read and interpreted by the cellular machinery to determine transcript stability, localization, translation efficiency, and decay kinetics. These features are not independent; they function as an integrated regulatory code in which modification status can influence poly(A) tail length, splice isoform choice can determine modification sites, and both can vary dynamically across cell types, developmental stages, and disease states.

Short-read RNA-seq fragments this integrated code into disconnected pieces, then relies on computational reconstruction to approximate the original transcript structures. cDNA-based long-read methods (PacBio Iso-Seq, ONT cDNA-seq) preserve the full-length connectivity through reverse transcription and PCR, but the amplification step destroys all modification information and introduces RT-dependent biases that systematically underrepresent GC-rich transcripts, structured RNA regions, and low-abundance isoforms. Direct RNA sequencing eliminates both of these information barriers by sequencing the native RNA molecule directly, preserving the complete transcript structure, modification landscape, and poly(A) tail status of every captured molecule. This makes dRNA-seq the only method capable of simultaneously resolving what the transcript is (sequence and splice isoforms), what state it is in (base modifications), and how it is regulated (poly(A) tail length) from a single experiment.

Our Nanopore Direct RNA Sequencing service makes this integrated transcriptome-epitranscriptome analysis accessible as a commercial CRO service, with validated RNA004 chemistry protocols, modified-base-aware bioinformatics pipelines, and flexible project configurations matched to eukaryotic transcriptome complexity.

Nanopore dRNA-seq Provides Five Data Types from a Single Sequencing Experiment — Sequence, Modifications, Poly(A) Tail, Isoforms, and Expression

Nanopore dRNA-seq Addresses Epitranscriptomics, Isoform Discovery, Poly(A) Tail Analysis, Cancer Biology, and Host-Pathogen Research

Beyond established applications in epitranscriptomics and isoform discovery, Nanopore direct RNA sequencing is increasingly deployed in clinical and translational settings — including rapid pathogen identification and antimicrobial resistance profiling directly from clinical samples, liquid biopsy-based detection of tumor-specific RNA modifications and fusion transcripts in circulating RNA, and real-time quality control of synthetic mRNA therapeutics where sequence fidelity, modification status, and poly(A) tail integrity are critical quality attributes that only native RNA sequencing can assess simultaneously from a single assay.

RNA Modification Mapping and Epitranscriptomics

• Single-nucleotide detection and quantification of m⁶A across the eukaryotic transcriptome, including identification of condition-specific differential modification sites, modification stoichiometry analysis, and co-occurrence of multiple modification types on the same transcript molecule
• Profiling of m⁵C, pseudouridine (Ψ), inosine (A-to-I editing), and 2′-O-methylation from the same dRNA-seq dataset using modified-base-aware basecalling models or comparative signal analysis against unmodified control libraries
• Integration of modification data with transcript isoform usage and expression level to investigate the regulatory cross-talk between epitranscriptomic marks and transcript fate — linking specific modification patterns to isoform-specific translation efficiency, RNA stability, or subcellular localization

Full-Length Native Transcript Isoform Discovery and Annotation

• Direct observation of complete transcript isoforms in native RNA, including alternative splice variants, alternative promoter usage, alternative polyadenylation (APA) isoforms with intact 3′ ends, and circular RNA back-splice junctions — without the isoform-level biases introduced by RT and PCR during cDNA library preparation
• Validation of computationally predicted isoforms from short-read data, providing definitive evidence for novel isoforms with full-length native transcript support that includes both splice architecture and 5′/3′ boundary information
• Single-molecule resolution of isoform heterogeneity within individual genes, including detection of mutually exclusive exon usage, intron retention events, and complex alternative splicing patterns that are ambiguous or invisible in short-read data

Poly(A) Tail Length Analysis and mRNA Regulation

• Transcript-resolved poly(A) tail length measurement at single-nucleotide resolution across the entire transcriptome, enabling direct analysis of tail length distributions for individual transcript isoforms and correlation with modification status on the same RNA molecules
• Dynamic poly(A) tail profiling across biological conditions — development, stress response, disease progression — revealing condition-specific changes in tail length that regulate mRNA stability and translational output at the transcript level
• Integration with PolyA Tail Length Analysis for focused studies of specific gene families, or combined analysis with modification data for comprehensive mRNA regulatory state characterization

Cancer and Disease Transcriptome Characterization

• Simultaneous detection of cancer-specific transcript isoforms, RNA modification dysregulation, and poly(A) tail abnormalities from tumor RNA samples using a single dRNA-seq experiment — providing integrated molecular profiles that correlate transcript structural changes with epitranscriptomic alterations in cancer pathways
• Detection of fusion transcripts and read-through transcription events with native RNA evidence, including modification status of fusion breakpoints and isoform-level expression of oncogenic fusion variants
• Linked analysis of m⁶A modification status and mRNA stability for oncogenes and tumor suppressors, leveraging the per-read modification and poly(A) tail data from dRNA-seq to investigate post-transcriptional regulatory mechanisms in tumorigenesis

Viral RNA and Host-Pathogen Transcriptomics

• Direct sequencing of viral RNA genomes and subgenomic RNAs without amplification bias, providing full-length viral transcript structures, modification landscapes, and poly(A) tail dynamics that are critical for understanding viral replication, immune evasion, and pathogenicity
• Simultaneous host and viral transcriptome analysis from infected samples, enabling direct observation of virus-induced changes in host RNA modification patterns, splice isoform usage, and poly(A) tail regulation during the infection time course

Our dRNA-seq Workflow Covers RNA Selection Through Signal-Based Basecalling and Multi-Modality Bioinformatics

1. RNA Selection and Quality Assessment

High-quality RNA is the single most important determinant of dRNA-seq success. Poly(A)-selected RNA (recommended for mRNA-focused studies) or total RNA (with optional rRNA depletion) is assessed by microfluidic electrophoresis. RNA integrity (RIN ≥ 8 recommended for standard dRNA-seq), concentration, and purity are established. DNase treatment is performed to eliminate genomic DNA contamination, which competes with RNA for adapter ligation and reduces sequencing yield. For total RNA samples or bacterial RNA (which lacks poly(A) tails), a controlled polyadenylation step is added before adapter ligation.

2. Adapter Ligation and Reverse Transcription

The ONT RNA004 adapter, which includes the motor protein docking site and a reverse transcription primer binding site, is ligated to the RNA 3′ end. Reverse transcription generates a cDNA strand that remains hybridized to the RNA, forming the RNA-cDNA duplex required for processive motor protein loading. Importantly, this RT step uses only a short primer extension — it is not a full cDNA synthesis reaction and does not involve PCR amplification. The cDNA serves exclusively as a structural component for nanopore translocation; the sequencing signal is derived from the native RNA strand.

Figure 1. Nanopore Direct RNA Sequencing workflow: from RNA extraction and quality assessment through RNA004 adapter ligation, reverse transcription, motor protein loading, and nanopore sequencing with simultaneous basecalling and modification detection.

3. Motor Protein Loading and Nanopore Translocation

The motor protein is loaded onto the RNA-cDNA duplex. Once loaded onto the PromethION or GridION flow cell, the motor protein processively unwinds the RNA-cDNA duplex and feeds the RNA strand through the nanopore in a 3′-to-5′ direction at approximately 70–100 nucleotides per second. As each nucleotide passes through the pore, it modulates the ionic current in a sequence- and modification-dependent manner. The current is measured at kHz frequency, generating a continuous signal trace that encodes the complete transcript information.

4. Real-Time Basecalling and Signal Processing

The ionic current signal is basecalled in real time using ONT Dorado with RNA-specific models that can distinguish modified bases from their unmodified counterparts. Simultaneously, the signal corresponding to the poly(A) tail region is analyzed for tail length estimation. The resulting data stream includes per-read nucleotide sequences, per-nucleotide modification probability scores, poly(A) tail length estimates, and per-read quality metrics — all derived from the same native RNA molecule during a single translocation event.

5. Multi-Modality Bioinformatics Analysis

Basecalled reads are aligned to the reference transcriptome or genome using splice-aware aligners (minimap2 with splice parameters). Downstream analysis includes: isoform detection and quantification (bambu, LIQA, or Flair), modification calling through comparative analysis against unmodified controls or using neural-network-based modification probability models (m⁶A via m6Anet, DRUMMER, or TandemMod; pseudouridine via NanoPsi or U-to-C error profiling; inosine via inosine-preferring alignment features), poly(A) tail length estimation (tailfindr or nanopolish), and integrated multi-modality correlation analysis linking modification status with isoform choice and poly(A) tail length on individual RNA molecules.

Sample Input Requirements for dRNA-seq — RNA Quality, Input Amount, and Flow Cell Configuration for Poly(A) and Total RNA Workflows

Case Study: Transfer Learning-Enabled Multi-Modification Detection by Nanopore Direct RNA Sequencing

Wu Y, Shao W, Yan M, et al. Transfer learning enables identification of multiple types of RNA modifications using nanopore direct RNA sequencing. Nature Communications. 2024;15:4049. (CC BY 4.0)

1. Background

Existing computational methods for detecting RNA modifications from nanopore direct RNA sequencing data are predominantly trained to recognize a single modification type (typically m⁶A) and require large, modification-specific training datasets. This single-modification paradigm limits the utility of DRS data, because a single dRNA-seq experiment simultaneously captures signal from all modification types present on the native RNA. The authors developed TandemMod, a transfer learning-based deep learning framework capable of identifying multiple RNA modification types (m⁶A, m⁵C, m⁷G, pseudouridine Ψ, and inosine) from a single nanopore DRS dataset, using an in vitro epitranscriptome (IVET) training strategy and transfer learning to reduce the training data requirements for each modification type.

2. Methods

In vitro epitranscriptome (IVET) datasets were generated from cDNA libraries containing thousands of transcripts with controlled modification labels for five modification types. These IVET libraries were sequenced by ONT direct RNA sequencing to produce training data with known modification ground truth. The TandemMod architecture uses a 1D convolutional neural network (CNN) with bidirectional long short-term memory (Bi-LSTM) layers and an attention mechanism, trained on the IVET datasets. Transfer learning was then applied to extend the model to each additional modification type, requiring substantially less training data than de novo model training. The framework was validated on human cell line DRS data and applied to rice transcriptome data from plants grown under normal and high-salt stress conditions.

3. Results

Fig. 2. TandemMod model and performance evaluation on published datasets. From Wu Y, Shao W, Yan M, et al. (2024, Nature Communications, CC BY 4.0).

Key Findings

• High-accuracy multi-modification detection from single DRS datasets — TandemMod achieved accurate detection of m⁶A and m⁵C modification sites validated on both in vitro transcripts and human cell line data, with significantly improved read-level and site-level performance compared to existing single-modification tools (nanom6A, m6Anet, tombo, xPore)
• Transfer learning reduces training data requirements — The transfer learning approach enabled extension to m⁷G, pseudouridine (Ψ), and inosine modification detection with substantially reduced training data sizes and computational requirements, demonstrating the feasibility of multi-modification analysis from standard DRS experiments without modification-specific training datasets for every target modification
• Cross-species applicability across plant and mammalian transcriptomes — TandemMod successfully identified modification sites in rice (Oryza sativa) DRS data, revealing m⁶A and m⁵C co-modification patterns on transcripts responsive to high-salt stress — demonstrating that DRS-based multi-modification analysis can be applied across diverse eukaryotic species without species-specific retraining
• IVET datasets as benchmark resources — The in vitro epitranscriptome (IVET) datasets generated for this study serve as reusable, publicly available benchmark resources for the development and validation of new modification detection methods, supporting continued improvement in DRS-based epitranscriptomics analysis
• Practical framework for multi-modality epitranscriptomics — This study established a practical computational framework that enables researchers to extract comprehensive modification information from standard dRNA-seq experiments, transforming a single DRS dataset from a single-modification profile into a multi-modality epitranscriptome resource

4. Conclusions

This study provides a validated computational framework for extracting multi-modification information from standard Nanopore direct RNA sequencing data, demonstrating that the information content of a single dRNA-seq experiment extends far beyond single-modification profiling. The TandemMod transfer learning approach establishes that comprehensive epitranscriptome characterization — covering m⁶A, m⁵C, m⁷G, Ψ, and inosine — is achievable from standard DRS libraries without modification-specific sample preparation. These findings validate our dRNA-seq service design: by deploying modified-base-aware basecalling and multi-modification analysis pipelines, we ensure that our dRNA-seq service delivers the full multi-modality information content that direct RNA sequencing is uniquely capable of providing.

When to Choose Nanopore Direct RNA Sequencing — and When Alternative Methods May Be More Suitable

Choose Nanopore direct RNA sequencing when:

• Native RNA modification detection is a primary research objective — If your project requires single-nucleotide, single-molecule modification information (m⁶A, m⁵C, Ψ, inosine), dRNA-seq is the only commercially available method that detects modifications from native RNA without antibodies, chemical conversion, or amplification. Every other RNA-seq method destroys modification information during RT or PCR
• Simultaneous multi-layer transcriptome analysis from a single experiment is valuable — dRNA-seq delivers transcript sequence, isoform structure, base modification status, poly(A) tail length, and expression quantification from the same native RNA reads — replacing separate RNA-seq, isoform-seq, MeRIP-seq, and poly(A) tail assays with a single sequencing experiment
• Poly(A) tail length information is required and must be linked to specific transcript isoforms — Only dRNA-seq measures poly(A) tail length at single-nucleotide resolution on individual RNA molecules while simultaneously identifying the transcript isoform and modification status — enabling correlation analyses that require the molecular linkage between these features
• You need full-length native transcript information without RT-PCR bias — For applications where amplification bias must be eliminated — including accurate quantification of GC-rich transcripts, detection of structured RNA regions, and characterization of native transcript ends — dRNA-seq avoids the RT-dependent drop-off and template-switching artifacts of cDNA-based methods
• RNA therapeutics or synthetic RNA quality control requires comprehensive characterization — dRNA-seq provides a single-assay QC platform for synthetic mRNA, assessing sequence fidelity, modification status, poly(A) tail integrity, and RNA length distribution simultaneously from a single sequencing run

Consider alternative methods when:

• High-throughput gene expression quantification is the primary goal — For projects focused on differential expression across many samples at the lowest per-sample cost, short-read RNA-seq (Illumina) provides 10–100× more reads per dollar and benefits from decades of method optimization and vast reference datasets. dRNA-seq throughput (~5–20 million reads/flow cell) is not optimized for high-depth expression quantification
• Deep isoform discovery across many samples is required — For projects requiring comprehensive transcript isoform catalogs from dozens of samples, cDNA-based long-read methods (ONT cDNA-seq or PacBio Iso-Seq) provide higher throughput and longer reads at lower cost per sample, at the cost of losing modification information
• Your RNA samples are degraded, limited in quantity, or derived from FFPE tissue — dRNA-seq requires high-quality, high-integrity RNA (RIN ≥ 8) and relatively high input amounts (~500 ng–1 µg polyA+ RNA). For degraded RNA, low-input samples, or FFPE material, short-read RNA-seq or cDNA-based long-read methods are more robust and practical
• Single-cell or very low-input transcriptomics is required — Current dRNA-seq protocols are not compatible with single-cell RNA-seq workflows. Single-cell transcriptomics projects should use standard scRNA-seq platforms; dRNA-seq can be applied to bulk RNA from FACS-sorted cell populations where required

CD Genomics provides free project consultation to help determine the optimal RNA sequencing strategy for your specific research questions. Contact our scientists to discuss your project requirements.

dRNA-seq Deliverables Include Basecalled Data, Isoform Annotations, Modification Profiles, and Integrated Multi-Modality Reports

1. Basecalled FASTQ files with per-read modification probability tags (MM/ML SAM tags) from Dorado v5+ RNA-specific modified-base models, ready for downstream analysis and public data deposition

2. Genome-aligned BAM files with splice-aware alignments, modification probability tracks, and read-level poly(A) tail length annotations, suitable for visualization in IGV or UCSC Genome Browser

3. Transcript isoform annotation and quantification report including full-length isoform structures, isoform-level expression matrices, splice junction analysis, and comparison with reference annotations

4. RNA modification analysis report with per-nucleotide modification probability tracks (bigWig), significant modification sites with genomic coordinates and confidence scores, condition-specific differential modification analysis, and per-read modification heterogeneity visualization

5. Poly(A) tail length analysis report with per-transcript poly(A) tail length distributions, transcript-resolved tail length summaries, and integrated correlation analysis linking modification status, poly(A) tail length, and isoform usage at single-molecule resolution