Direct RNA sequencing with nanopore technology reads native RNA molecules one at a time, without first converting them to DNA or making copies. A single-stranded RNA molecule is fed through a tiny protein pore embedded in a membrane, and as each stretch of the molecule passes through, it creates a distinct disruption in an electrical current flowing across that membrane. A computer then translates those current disruptions into a sequence of nucleotide bases. The result is a readout of the original RNA as it existed inside the cell, complete with chemical modifications, full-length transcript structure, and poly(A) tail lengths that other sequencing methods typically erase or distort during sample preparation.
How the Nanopore Reads an RNA Strand
The core of the technology is a nanoscale protein channel sitting in a synthetic membrane. An ionic current flows through the pore. When an RNA molecule is loaded, a motor protein latches onto one end and ratchets it through the pore one segment at a time. As each short stretch of RNA occupies the narrowest part of the channel, it partially blocks the current in a pattern that depends on which bases are present. The sensor records these current shifts continuously, and software called a basecaller converts the raw electrical signal into a sequence of A, C, G, and U bases.1PubMed Central. Direct Sequencing of RNA and RNA Modification Identification Using Nanopore
What makes this “direct” is what it skips. Conventional RNA sequencing methods reverse-transcribe RNA into complementary DNA, then amplify that DNA millions of times before reading it. Both of those steps introduce artifacts. Reverse transcription can stall or skip over structured regions of the RNA. Amplification creates copies that may not faithfully represent the original proportions of different transcripts. Nanopore direct RNA sequencing bypasses both steps entirely, reading the RNA molecule as-is in real time.2PubMed Central. Highly parallel direct RNA sequencing on an array of nanopores
Why Skipping the Copy Step Matters
When RNA carries chemical modifications, those modifications are biologically meaningful. A methyl group added to one base can change how a protein is made, how long the RNA survives in the cell, or how efficiently it gets translated. More than 170 types of RNA modifications have been catalogued, and they play roles in everything from brain development to cancer progression. Conventional sequencing erases most of this information during the reverse-transcription step because the enzyme that copies RNA into DNA does not faithfully reproduce every modification. Some modifications cause the enzyme to misread a base; others are simply ignored.
Direct RNA sequencing preserves these modifications in the signal. Because the actual modified nucleotide passes through the pore, its distinctive current pattern can be detected computationally. Researchers have developed machine-learning tools that can simultaneously profile two of the most common modifications in human messenger RNA from a single nanopore run.3PubMed Central. Simultaneous nanopore profiling of mRNA m6A and pseudouridine reveals translation coordination Another modification, pseudouridine, leaves a telltale signature in the basecaller output: the software systematically misreads the modified U as a C, producing a mismatch error at known pseudouridine sites with rates that can exceed 90% at some positions.4Nature Communications. Semi-quantitative detection of pseudouridine modifications and type I/II hypermodifications in human mRNAs using direct long-read sequencing That consistent error pattern, counterintuitively, becomes a useful signal for mapping where the modification occurs.
Reading Full-Length Transcripts and Their Variants
Genes do not produce just one type of RNA. Through a process called alternative splicing, a single gene can generate multiple transcript variants, or isoforms, each potentially coding for a slightly different protein or regulated in a different way. Short-read sequencing chops RNA into tiny fragments and then tries to computationally reassemble which isoform each fragment came from, a process that often fails for complex genes with many overlapping variants.
Because nanopore sequencing reads the entire RNA molecule from end to end, it captures the full isoform structure in a single read. A study using direct RNA sequencing in the roundworm C. elegans detected over 14,000 isoforms, including 813 novel isoforms from 782 genes that were absent from the existing reference annotation. The same dataset revealed 617 previously unknown 3′ untranslated regions.5Nucleic Acids Research. Full-length direct RNA sequencing reveals extensive remodeling of RNA expression, processing and modification in aging Caenorhabditis elegans These are the kinds of features that short-read methods routinely miss, because reconstructing a full transcript from 150-base fragments is fundamentally ambiguous when multiple isoforms share large stretches of identical sequence.
The technology also measures poly(A) tail lengths on individual transcripts. The poly(A) tail is a string of adenine bases added to the end of most messenger RNAs, and its length influences how long the RNA survives and how efficiently it is translated into protein. Dedicated software tools can estimate these tail lengths directly from the raw nanopore signal without needing to align the reads to a reference genome first.6PubMed Central. tailfindr: alignment-free poly(A) length measurement for Oxford Nanopore RNA and DNA sequencing Getting this measurement on a per-molecule basis is something conventional sequencing simply cannot do.
Accuracy and Where the Errors Show Up
Direct RNA sequencing is noisier than DNA nanopore sequencing, and both are noisier per-read than short-read platforms like Illumina. The error profile has a distinctive pattern. In nanopore DNA sequencing, long stretches of repeated single nucleotides (homopolymers) are the main source of trouble. In direct RNA sequencing, the errors are more evenly spread: short homopolymers of just two or three bases and mixed-base stretches (heteropolymers) together account for the vast majority of errors. These motifs make up roughly 91.5% of the sequenced data in a human transcriptome and show similar rates of deletions and mismatches.7PubMed Central. Sequencing accuracy and systematic errors of nanopore direct RNA sequencing
Among homopolymers specifically, accuracy drops as the repeated stretch gets longer, and the basecaller tends to undercount the number of bases. Homopolymers of A perform best, remaining above 50% accuracy even at a length of five. Stretches of C and G are harder for the system to resolve accurately.8bioRxiv. Sequencing accuracy and systematic errors in nanopore direct RNA sequencing These systematic error patterns matter because they can be confused with genuine biological variation or modification signals if not handled carefully in downstream analysis.
Improved basecalling software is helping close the accuracy gap. A recently developed basecaller called Coral achieved up to a 6.17% improvement in accuracy on human RNA samples compared to the manufacturer’s own Dorado basecaller. That accuracy gain translated directly into practical benefits: 26% more annotated transcript isoforms were detected, and when the data was used to distinguish which parental copy of a chromosome an RNA came from (haplotype phasing), errors dropped by more than 75%.9Nature Communications. A dual context-aware basecaller for nanopore direct RNA sequencing
Counting Transcripts Reliably
Beyond just identifying what transcripts are present, researchers often want to know how much of each one there is. Gene expression quantification is a bread-and-butter application in biology. Direct RNA sequencing, because each read represents one original molecule, should in principle give a straightforward count. In practice, the lower throughput of nanopore sequencing compared to short-read platforms means that rare transcripts may not get enough reads to quantify accurately.
Benchmarking with synthetic RNA standards (called sequins) has shown that measured abundance correlates well with known abundance at both the gene and isoform level, with correlation values of 0.96 and 0.90 respectively. Below a certain concentration threshold, though, detection becomes unreliable due to limited sequencing depth.10PubMed Central. Accurate expression quantification from nanopore direct RNA sequencing with NanoCount A systematic comparison of long-read RNA sequencing methods found that transcript quantification with long-read tools generally lags behind short-read tools, a gap driven primarily by throughput and error limitations.11Nature Methods. Systematic assessment of long-read RNA-seq methods for transcript identification and quantification So direct RNA sequencing excels at telling you what isoforms exist and what modifications they carry, but if your main goal is precisely measuring the abundance of thousands of transcripts at once, short-read methods still have a throughput advantage.
Practical Considerations for Sample Quality
Because the technology reads native RNA, the quality of the input material matters a lot. RNA is inherently fragile and degrades quickly if not handled properly. Degraded RNA produces truncated reads that can bias results toward the 3′ end of transcripts (the end where sequencing begins, near the poly(A) tail) and away from the 5′ end. Research into this problem found that samples with a high RNA integrity number (above 9.5) behave as essentially undegraded. Samples with integrity numbers above 7 can still be used, but the data requires computational correction to account for degradation artifacts.12PubMed Central. Pervasive effects of RNA degradation on Nanopore direct RNA sequencing
This sensitivity to RNA quality is a practical constraint. Cell lines grown under controlled conditions often yield beautifully intact RNA, but clinical tissue samples, archived specimens, or fieldwork samples may not. Anyone planning a direct RNA sequencing experiment needs to prioritize careful RNA extraction and storage, or accept the trade-offs of working with partially degraded material.
Applications in Virology and Vaccine Quality
Viral RNA genomes are a natural fit for this technology. Many RNA viruses have relatively small genomes that can be captured in full-length reads, and direct sequencing lets researchers examine untranslated regions, splice variants, and base modifications without the biases introduced by reverse transcription and PCR. Work on influenza A virus demonstrated that nanopore direct RNA sequencing could capture the coding-complete viral genome and examine its untranslated regions in their native form, including the potential to detect base modifications and transcriptional changes directly.13Scientific Reports. Direct RNA Sequencing of the Coding Complete Influenza A Virus Genome
The technology has also found a role in quality control for mRNA vaccines. During vaccine manufacturing, it matters whether the synthetic mRNA has the correct sequence, the right length, and the intended chemical modifications. The modified nucleoside used in COVID-19 mRNA vaccines, N1-methylpseudouridine, produces a characteristic basecalling error when read by nanopore: the software misclassifies it as cytosine about 62% of the time, with the remaining calls split to uridine. This predictable error signature allows researchers to verify that the modification is present throughout the vaccine molecule. A sequencing-based quality analysis framework demonstrated that direct RNA sequencing could comprehensively measure key mRNA vaccine quality attributes, including sequence, length, integrity, and purity.14PubMed Central. mRNA vaccine quality analysis using RNA sequencing
Going Beyond Messenger RNA
The standard direct RNA sequencing protocol is designed for polyadenylated RNA, meaning it captures the poly(A) tail to initiate sequencing. Most messenger RNAs have poly(A) tails, but many important non-coding RNAs do not. Ribosomal RNA, transfer RNA, small nuclear RNA, and various regulatory RNAs lack poly(A) tails and are therefore invisible to the standard workflow.
Researchers have worked around this by designing custom adapters that ligate directly to non-polyadenylated targets. In one proof-of-concept, targeted sequencing of the small nuclear RNA 7SK achieved high coverage and enabled detailed mapping of modifications across the molecule, including sites dependent on a specific methyltransferase enzyme.15Nature Communications. RNA modifications detection by comparative Nanopore direct RNA sequencing Another approach used capture probes to enrich specific RNA species before sequencing. Probes targeting the MYCN oncogene achieved a purification factor of nearly 5,000-fold, with 65% of the resulting mapped reads aligning to the target gene’s transcripts.16PubMed Central. A gene-specific RNA enrichment protocol for nanopore direct-RNA sequencing These enrichment strategies are valuable when you care deeply about one gene or a small panel and want the modification and isoform detail that direct RNA sequencing provides, without needing to sequence the entire transcriptome.
Detecting RNA Secondary Structure During Sequencing
RNA is not just a linear string of bases. It folds back on itself to form hairpins, loops, and other secondary structures that influence how the molecule functions. Conventionally, studying RNA structure requires separate experiments, often involving chemical probing or enzymatic digestion. Researchers discovered that the motor protein controlling RNA translocation through the nanopore can actually sense secondary structures about 11 to 12 nucleotides ahead of where it sits. When the motor encounters a strong structure downstream, its stepping dynamics change in a detectable way. This means that sequence and structure information can, in principle, be extracted from the same nanopore run simultaneously, without any additional chemical treatment or conversion to cDNA.17bioRxiv. Secondary Structure Detection Through Direct Nanopore RNA Sequencing
This capability is still early-stage, and extracting clean structural signals from noisy current traces is computationally challenging. But the idea that a single molecule passing through a pore could yield its sequence, its modifications, and its folding state is remarkably ambitious, and the initial results suggest it is at least partially achievable.
The Shift from RNA002 to RNA004 Chemistry
Oxford Nanopore Technologies, the company behind the platform, recently updated its direct RNA sequencing chemistry from the RNA002 to the RNA004 kit. The new chemistry comes paired with an improved basecaller and includes built-in models for detecting four types of RNA modifications directly, without the need for third-party tools or matched control samples.18PubMed Central. Evaluation of Nanopore direct RNA sequencing updates for modification detection Independent testing with viral RNA confirmed that the new chemistry produces better read quality and longer reads compared to its predecessor.19PubMed. Comparison of direct RNA sequencing of Orthoavulavirus javaense using two different chemistries on the MinION platform
The upgrade comes with a practical trade-off, though. The manufacturer does not offer a Flongle flow cell for the RNA004 chemistry. The Flongle is a small, inexpensive flow cell that was popular for low-input or pilot experiments. Without it, every sample must run on a larger MinION or PromethION flow cell, which increases both hands-on time and cost per sample. One research group addressed this by developing a multiplexing strategy that allows several barcoded samples to share a single flow cell, bringing the per-sample cost back below what a Flongle run would have cost while actually increasing the data output per sample.20Nature Communications. Demultiplexing and barcode-specific adaptive sampling for nanopore direct RNA sequencing
Portability and Where the Sequencing Happens
One of the more unusual features of nanopore sequencing in general is that the hardware is small enough to carry in a backpack. The MinION device, roughly the size of a stapler, plugs into a laptop via USB. This portability has already transformed DNA sequencing in fieldwork settings, from tracking Ebola outbreaks in remote clinics to monitoring biodiversity in rainforests. Direct RNA sequencing inherits this portability, though the sensitivity of RNA to degradation means that field applications require careful cold-chain management of samples or very rapid processing after collection.21PubMed Central. Portable nanopore-sequencing technology: Trends in development and applications
The real-time nature of the data is another practical advantage. Because reads are generated as molecules pass through the pores, you can start analyzing results within minutes of beginning a run. For applications like pathogen surveillance, where a rapid answer matters more than completeness, this is a genuine benefit over batch-processing platforms that require the entire run to finish before any data is available. Adaptive sampling, a feature that selectively ejects molecules from the pore based on their identity in real time, further extends this flexibility by allowing the sequencer to focus its limited capacity on molecules of interest and reject unwanted ones.
When Direct RNA Sequencing Is and Is Not the Right Tool
The technology shines in situations where preserving information about the native RNA molecule is the primary goal. If you need to know which chemical modifications a transcript carries, how long its poly(A) tail is, what its full-length isoform structure looks like, or whether its secondary structure is intact, direct RNA sequencing provides information that no other method currently can from a single experiment. For mRNA vaccine quality control, viral transcript mapping, and epitranscriptomic profiling, it occupies a niche that competing platforms cannot easily fill.
It is less well-suited when the priority is sheer throughput, ultra-low error rates, or quantification of very rare transcripts. Short-read RNA sequencing still generates far more reads per dollar, and its per-base error rate, after consensus, is lower. For standard differential gene expression experiments comparing dozens of samples across thousands of genes, short-read platforms remain the workhorse. Direct RNA sequencing is not a wholesale replacement for those workflows; it is a complementary technology that answers different kinds of questions. The field is still actively developing the computational tools, library preparation protocols, and chemistry improvements that could widen its applicability over time.