Successful lab-in-the-loop approaches require that the design, build, and test phases are optimized both in terms of their individual processes and how they interact with one another. For example, many design approaches often treat manufacturing as a black box: A sequence is designed and later an RNA appears. This abstraction ignores a critical reality: Not all RNA sequences are equally manufacturable, and manufacturing problems have cascading consequences that corrupt downstream data quality. In this eBlog we review some of the common challenges that can arise with RNA manufacturing and how to overcome them.

DNA fragment synthesis and assembly

The first step in manufacturing custom RNA constructs involves synthesizing DNA fragments that encode your design and assembling them into plasmids for bacterial cloning. However, not every sequence can be effectively synthesized. Sequences with high GC content, repetitive elements, or strong secondary structures can increase error rates during chemical synthesis and can be challenging to assemble into full-length constructs.

If the sequence designs are not screened prior to synthesis, the issues may not be discovered until well into the manufacturing process, often when transformation plates yield no colonies or sequencing reveals incorrect assemblies. For programs trying to evaluate panels of 10-20 candidates, a 20-30% failure rate at this step increases both the costs and timelines.

Eclipsebio addresses these issues through upfront manufacturability assessment and vendor diversity. Our comprehensive design review process flags sequences with predicted assembly risks, problematic motifs, or cloning challenges before making DNA fragments into plasmids.

Plasmid manufacturing and bacterial transformation

Once DNA fragments are assembled into plasmids, the next step is bacterial transformation and clone selection. Specific designs may form toxic peptides in this step, which pressure the bacteria to mutate the plasmid or truncate problematic regions, including homopolymeric stretches such as poly(A) tails. These mutations often aren’t discovered until after plasmid purification and sequencing, typically 2-3 days post-transformation. Mutations and problematic regions can negatively impact the entire manufacturing process, especially as drug developers scale up. This lengthens the production timeline and increases cost.

Even when plasmid identity remains intact, these design flaws lower plasmid prep yields while endotoxin levels remain fixed or even elevated. This directly increases the immunogenicity of the input material to the IVT reaction.

At Eclipsebio we use full-length plasmid nanopore-based sequencing at the clone screening stage to confirm 100% sequence identity before scale-up. In addition, we verify poly(A) tail length via Sanger sequencing to ensure no truncation has occurred. We also maintain clone libraries and utilize our design experience and datasets to navigate challenging sequences. The goal isn’t just to produce a plasmid; it’s to produce a verified plasmid that won’t introduce additional variables into your downstream experiments once your process moves to scale.

In vitro transcription

The final manufacturing step, in vitro transcription (IVT), converts linearized plasmid templates into RNA. Here, sequence-dependent effects on transcription efficiency, RNA secondary structure during synthesis, and susceptibility to degradation can reduce yield and percent full-length product. These issues force developers to scale up reaction volumes, using extra DNA that is potentially contaminated with residual endotoxin, reaction proteins, and dsRNA that can carry into the drug substance.

To overcome these challenges, Eclipsebio’s IVT workflow includes process selection based on application requirements (precipitation-based purification for standard research applications, oligo-dt based purification for primary cells or in vivo work) and fragment analyzer assessment of integrity to ensure percent of full-length product (%FLP) is greater than 75% ( >80-90% for most constructs). We also provide advanced drug substance characterization through our eMERGE analytics platform to obtain actionable insights into different dimensions of RNA quality, including dsRNA levels and RNA integrity.

Quality control as experimental variable isolation

In every step of manufacturing, quality control is essential. By incorporating analytics at each stage, developers can better distinguish if issues are related to their sequence or artifacts in the manufacturing process.

As an example, consider a common scenario: Two candidate mRNA constructs (Candidate A and Candidate B) are developed that differ in UTR structure. Both sequences are sent to a contract manufacturer, and the Certificates of Analysis from both show acceptable concentration and purity.

During subsequent testing, Candidate A shows 60% of Candidate B’s expression level and produces a stronger innate immune response. A reasonable conclusion is that Candidate B’s UTR design is superior, and it advances to the next round of optimization.

However, there are several unmeasured attributes that could show the differences are due to manufacturing rather than UTR efficacy:

• Candidate A’s plasmid underwent poly(A) tail truncation during bacterial culture. The shortened poly(A) tail reduced translational efficiency and RNA stability, confounding your expression comparison.
• Candidate A had 78% FLP while Candidate B had 91% FLP. (Both passed the manufacturer’s “>70% FLP” spec, so it wasn’t flagged.) This would then lead to changes in protein production and immunogenic responses.
• In this scenario, you’re not comparing UTR designs, you’re instead comparing manufacturing runs. The conclusion that Candidate B is superior might be correct, or it might be an artifact of differential manufacturing quality.

This is the core problem with treating manufacturing as a black box: Without rigorous quality control at each process step, you cannot isolate sequence-dependent effects from manufacturing-dependent effects.

At Eclipsebio our multidimensional quality control approach is designed for variable isolation:

• Sequence verification: Oxford Nanopore sequencing of plasmid clones confirms 100% identity to design across all functional regions (5′ UTR, ORF, 3′ UTR) and verifies poly(A) tail integrity before scale-up. You know you’re testing the sequence you designed, not a mutated variant.
• Integrity assessment: Fragment Analyzer measurement of %FLP ensures material meets specifications (>75% minimum, typically >80-90%). Truncated products are quantified, not hidden.
• Purity characterization: Spectrophotometry (A260/A280, A260/A230) provides baseline assessment of contaminants. Advanced quality control with our eMERGE analytics platform provides detailed insights into other aspects of RNA quality, capturing details that are missed by other approaches such as changes in dsRNA content between batches.
• The result is that when you test Candidate A versus Candidate B, you’re making a controlled comparison. Differences in expression or immunogenicity can be attributed to sequence features with confidence because manufacturing variables have been measured and minimized.

Manufacturing and design knowledge compounds over iterations

Through manufacturing hundreds of constructs, troubleshooting challenges, and recognizing patterns that predict problems before synthesis, Eclipsebio has built a knowledge base for RNA design. We use this knowledge throughout our manufacturing process, especially in our upfront sequence assessment.

This knowledge allows us to complete the manufacturing cycle quickly, generating a greater number of high-quality mRNA sequences for functional RNA-based therapeutics.

Manufacturing-informed design allows for:

• Rapid prototyping with data confidence: Since we know that every construct meets sequence identity, integrity, and purity specifications, we can attribute differences in functional readouts between constructs to sequence features, not manufacturing artifacts.
• Assay readout clarity: Our advanced quality analysis assays quantify residual contaminants including dsRNA and proteins, enabling us to interpret results with confidence that unexpected immune activation or toxicity isn’t due to manufacturing carryover.
• Exploration of design space: Knowing where the boundaries of manufacturability lay allows us to further explore the sequence space, gaining a broader understanding of their RNA’s effects.

Eclipsebio’s approach integrates manufacturability expertise at the design stage, implements QC gates throughout the build process, and provides full traceability for every batch. The result is research-grade RNA that enables confident iteration to know what is tested, attribute differences between constructs to sequence, and inform new designs with data from each cycle.

Interested in how manufacturing-informed design can accelerate your program? Contact Eclipsebio today.

The mRNA Innovators Award offers matching funds to support early-stage RNA drug developers* in the design and validation of a novel mRNA-based therapeutic using Eclipsebio’s mRNA drug discovery platform. 

As an awardee, our RNA therapeutics success team will design, prototype, and validate your therapeutic of interest using our design models and trusted RNA characterization methods. You will receive an aliquot of the mRNA for additional testing. In addition, you will maintain full ownership of the developed therapeutic with no requirements for licensing or royalties.  

With this award, Eclipsebio furthers its commitment to enabling high-impact RNA research and providing early-stage drug developers with the same RNA design and characterization assays trusted by established therapeutic developers. 

Interested early-stage RNA drug developers should fill out the form accessed from the button below. You will be asked in the application to submit a brief description of your therapeutic goals. If you pass the initial review, you will meet with our design team to review your project in more detail before the final selection. 

For any questions about the mRNA Innovators Award, application process, or how Eclipsebio supports our partners in their research and drug discovery programs, you can contact us at [email protected]. 

*To support early-stage innovation, this competition is only available to academic researchers or companies with pre-seed or seed-stage funding. 

Step 1: Design 

Designing a sequence is the first step of developing any RNA therapeutic, and prioritizing a high-quality design at the beginning of therapeutic development will make the rest of the process easier. Developers can design RNA sequences based on data from previous designs using advanced learning models. 

Terrain Bio, now part of Eclipsebio, has developed advanced learning models to design high-quality RNA sequences. The pairing of these learning models with Eclipsebio’s eVERSE AI-ready datasets, enables biopharmas and biotechs to obtain RNA designs that outperform existing approaches for both potency and stability.

Step 2: Make 

After designing the RNA sequence, it still needs to be tested in the real world. With our rapid prototyping capabilities, we can quickly manufacture the novel RNA sequences for robust validation with our eMERGE analytics platform. Our proprietary manufacturing protocols enable the synthesis of technically challenging RNAs, including linear mRNAs longer than 5 kb.

In addition to our own manufacturing capabilities for preclinical prototyping, we can also help you identify the ideal scale-up vendor from our global network of manufacturing partners.

Step 3: Test 

After manufacturing the initial RNA prototypes, they then need to be tested to identify the optimal candidate to move forward with IND filings.  We pair our AI-powered designs with empirical, sequencing-based analytics to provide comprehensive evaluations of each drug candidate through our eMERGE platform. Example insights from our platform include the determination of dsRNA sources, assessment of RNA secondary structure,  and measurements of ribosome dynamics. Included with each candidate validation is a comprehensive insights report to help you interpret the data and make actionable decisions for how to proceed.

Repeating the cycle to teach AI 

Designing, prototyping, and validating RNA are all important processes in themselves, but they are most powerful when used together. Our unified approach enables us to gather data for every step of the Design, Make, Test cycle enabling an active lab-in-the-loop approach for improving initial RNA sequence designs. 

For example, imagine a developer has designed an RNA sequence based on available AI data, which they then prototype and validate. During analysis, they find that the RNA is prone to hydrolysis and the RNA’s secondary structure allows for dsRNA formation in several regions. We feed this data back to our partner-specific AI models to guide improvements to increase therapeutic efficacy.

Design, Make, Test cycle at Eclipsebio 

At Eclipsebio, we complete this entire process in-house enabling the development of clinic-ready drugs on accelerated timelines. Our AI-guided design capabilities create RNA sequences optimized for manufacturing and efficacy, and our sequencing-based analytics validate manufactured sequences.

Ready to discover how our unified platform can accelerate and derisk your drug development programs? Contact Eclipsebio to get started. 

Want to learn more about Eclipsebio’s acquisition of Terrain Bio? Read the press release. 

SAN DIEGO, January 26, 2026 — Eclipse Bioinnovations, Inc. (Eclipsebio), the leader in sequencing-based analytics for RNA therapeutics, today announced the acquisition of Terrain Bio, a techbio company specializing in AI/ML RNA design and manufacturing analytics. The acquisition accelerates Eclipsebio’s strategy to deliver an integrated, data-first platform to support RNA therapeutic design, manufacturing, and validation.

Terrain Bio has developed advanced machine learning models for RNA sequence optimization and active learning workflows, alongside a best-in-class R&D-scale mRNA manufacturing platform, to connect computational design directly with experimental feedback. By integrating these capabilities with Eclipsebio’s established sequencing-based datasets and quality control platforms, including eMERGE^TM and eVERSE^TM, Eclipsebio will offer partners a unified Design, Make, Test solution that continuously improves as new data are generated in the development of RNA-based medicines.

“This acquisition meaningfully advances our vision for Eclipsebio,” said Peter Chu, CEO of Eclipsebio. “Terrain Bio’s proven AI design capabilities strongly complement our sequencing-first analytics, allowing us to support RNA drug developers earlier in development while continuing to deliver the deep sequencing-based validation our partners rely on for confident decision-making.”

The combined platform will enable RNA therapeutic developers to:

• Iteratively design optimized RNA constructs using AI modeling and active learning;
• Measure structure, translation, and impurities with high-resolution sequencing analytics; and
• Gain actionable insight into manufacturing robustness and quality-by-design.

“Eclipsebio’s deep expertise in sequencing-based validation and curated data repository makes this combination uniquely powerful,” said Chetan Tadvalkar, CEO of Terrain Bio. “Together, we close the gap between computational design and real-world experimental validation, helping RNA therapeutics reach the clinic faster and with greater confidence.”

Eclipsebio will continue to support existing Terrain Bio customers and will work closely with their partners to ensure a seamless transition.

About Eclipsebio

Eclipsebio is a private biotechnology company headquartered in San Diego developing first-in-class technologies, analytics, and platforms for the development of tomorrow’s RNA-based and RNA-targeting therapies. With the company’s extensive experience in supporting early-stage basic research through to evaluating preclinical therapies, Eclipsebio provides unparalleled support for obtaining deep insights into RNA and therapeutic biology. The company offers its solutions as end-to-end partnerships for biopharma and project-based services for academics. For more information about Eclipsebio, visit www.eclipsebio.com.

About Terrain Bio

As an AI-first mRNA CRO, Terrain Bio has developed a technology stack that explores a vast design space of RNA sequences to identify optimized candidates for expression, durability, and function. This tech stack is paired with a wet-lab platform capable of rapidly manufacturing and testing RNA constructs at quality levels that meet or exceed industry benchmarks. Together, these capabilities allow partners to efficiently design, build, and evaluate RNA sequences, accelerating progression toward viable drug candidates and enabling faster IND submissions.

Contacts

For media: [email protected]

For partnerships and platform inquiries: [email protected]

While these innovations are promising, developers must ensure their drug products meet regulatory standards. With IVT RNA, maintaining the integrity of the RNA is especially important to avoid impurities. Drug regulators are continuously improving their guidelines, and methods to check quality and integrity are evolving alongside these regulatory standards. 

Why IVT RNA integrity matters 

RNA is an inherently unstable molecule and will lose integrity over time. Temperature, pH, and length of the RNA all affect integrity, leading to denaturing or fragmentation under certain conditions. One prominent example is hydrolysis of the backbone, which can be exacerbated by pH changes or the presence of specific ions. In addition, RNase contamination can cleave RNA. The RNA must be a full, intact strand to make the correct protein for a therapeutic, so any breakage makes the therapeutic less effective.  

These factors are an issue for RNA therapeutics across the therapeutic lifecycle from development, to storage, to delivery. After a drug is developed, it must be stored at a high quality until it is administered, typically requiring the use of cold storage. Even after it is administered, the RNA must remain protected inside a delivery vehicle until it is taken up by cells and released into the cytoplasm. 

Standard methods to measure integrity 

Due to the critical importance of integrity in therapeutic efficacy, regulators have established strong guidelines for the amount of intact RNA that is required for release. Currently, most regulators recommend using well-established technologies such as capillary or agarose gel electrophoresis and liquid chromatography.  

One of the most common methods used to evaluate integrity is capillary gel electrophoresis. In this procedure, RNA is injected into a gel-filled capillary tube. RNA molecules then separate by size, with short fragments moving at different rates than full intact molecules. The size of the fragments present can be determined in comparison to a reference ladder. Although powerful, this approach is limited to only revealing the size of contaminating fragments. It can’t reveal the identity of the fragmented species or identify breakage hotspots. 

Sequencing for integrity 

Next-generation sequencing is an innovative technology that is increasingly used to gain insights into IVT RNA integrity and purity. For example, sequencing can answer regulatory questions about the 5’ cap and 3’ tail, determining the length of the cap and poly(A) tail. It can also identify the RNA’s sequence at a single-nucleotide resolution, including the identification of fragmented molecules. 

Nanopore sequencing is one approach that has been applied to identify fragmented RNAs. Since nanopore sequencing is long-read sequencing, it reads the whole RNA at once rather than fragmenting it as is done in short-read sequencing. This lets developers characterize the entire RNA, including the identification of specific fragments that are present as seen in work by Gunter et al. in Nature Communications.  

Integrity characterization at Eclipsebio 

At Eclipsebio, we offer innovative sequencing assays that help developers meet regulatory requirements. For example, eSENSE dsRNA locates dsRNA in an IVT RNA to show where IVT runoff is occurring along a sequence. This assay also measures how much dsRNA is present in an IVT RNA strand. When combined with an RNA structure identification assay, such as Eclipsebio’s eSHAPE assay, eSENSE dsRNA can help identify RNA secondary structures that are more prone to dsRNA formation, offering insights for developers to improve their therapeutic’s purity. 

Specifically for integrity, Eclipsebio’s nanopore sequencing approach uses Oxford Nanopore’s technology to directly read an entire RNA sequence. This long-read sequencing can find the length of a poly(A) tail and profile an entire RNA transcript on a single-nucleotide level, allowing for investigations of RNA integrity. 

Using sequencing to achieve regulatory success 

As RNA therapeutics advance, so do regulatory requirements. Methods to measure IVT RNA quality such as eSENSE dsRNA and nanopore sequencing help drug developers keep up with changing regulatory guidelines and create stable, clean, and effective RNA therapeutics. 

Interested in how Eclipsebio can help you validate the integrity of your IVT RNA? Reach out to us to get started. 

References

As we reach the end of 2025, we are looking back at some of the major moments in RNA science and innovation throughout the year. From breakthroughs in clinical trials, to the rise of AI models for RNA research, to the first successful personalized gene editing treatment, 2025 was a stellar year for RNA. We are excited to share some of the year’s most notable highlights in this eBlog. 

Continued advancements in RNA medicines and therapeutics

This year saw numerous advancements in developing RNA medicines and therapeutics. Here are just a few of these exciting advancements. 

• Moderna: The FDA approved Moderna’s RSV vaccine, mRESVIA, for adults 18–59 who are at increased risk for respiratory syncytial virus (RSV). The vaccine became available to this group, alongside those 60 years and over, this respiratory viral season. 
• HDT-Bio: The FDA cleared HDT-Bio’s Investigational New Drug application for the company’s HDT-321 vaccine against Crimean-Congo Hemorrhagic Fever. This is a significant step towards protecting public health from tick-borne illness. 
• Baby KJ: The first successful personalized gene editing treatment occurred at the Children’s Hospital of Philadelphia. Over several months, physicians gave the infant KJ, who was diagnosed with severe CPS1 deficiency, base editing therapy via lipid nanoparticles. Under close watch, baby KJ has not faced any major side effects and is healthily growing. 

Improvements in the stability of mRNA

One of the biggest challenges of mRNA drugs is keeping the drug substance stable from synthesis to protein production. Fortunately, there was a lot of research this year into the issue. 

• Ma et al. explored how optimizing mRNA sequences can improve stability, finding that adding AU-rich elements to the 3’ UTR made mRNA more stable. 
• Snell et al. discovered that the RNA-binding proteins (RBPs) ZC3H7A and ZC3H7B preferentially bind to non-optimal A/U3 codons, which makes the mRNA less stable. Depleting these RBPs reduces the expression of non-optimal codons, improving stability. 
• In their review, O’Leary et al. explained that circular RNA (circRNA) is more stable and has a longer half-life than linear mRNA. These characteristics mean that circRNA-encoded antigens could have a longer and more pronounced protein expression in vaccines. 
• Cao et al. reviewed many chemical modifications and delivery options for mRNA stability. One consideration was that scientists may be able to use less self-amplifying RNA (saRNA) than the required amount of mRNA to achieve the same or even more protein expression. 
• Wang et al. discussed how chemical modifications can make different types of RNA more stable. They proposed that alterations to the backbone and sugar of small RNAs, base changes to mRNA and other translatable RNAs, and backbone modifications to guide RNA can help improve RNA stability. 

Increased prominence of AI in RNA research

The use and accessibility of AI models for RNA therapeutics expanded throughout the year. From locating where miRNAs bind at a single-nucleotide scale to determining RNA secondary structures, AI can be used in almost every aspect of creating RNA medicines. 

• Li et al. developed mRNA-LM, a small language model that sequences the entire mRNA sequence, including UTRs and CDS. They used three separate models for UTRs and coding sequences, leading to an effective model for predicting translation and stability. 
• Qu et al. developed CRISPR-GPT, a large language model to automate and improve every step of CRISPR gene editing, from experiment designing to data analysis. 
• Kanuparthi et al. developed REPRESS, a deep learning model that can identify non-canonical miRNA binding sites and predict mRNA degradation. They trained REPRESS with miRNA binding data from miR-eCLIP. 

Innovation highlights at Eclipsebio

• This year we launched eVERSE, the first comprehensive portfolio of multidimensional RNA datasets for AI model training. Using Eclipsebio sequencing services, eVERSE offers high-quality datasets to better develop RNA therapeutics and train AI models. 
• We began offering nanopore sequencing this year, providing our partners with long-read data of their RNAs to measure the length of poly(A) tails and characterize RNA fragments. 

Looking forward to 2026

As the field of RNA therapeutics and medicines continues to advance, here are some research topics we predict to grow. 

• The use of AI in RNA drug development will continue to increase. As RNA datasets to train AI, such as eVERSE, become more accessible, researchers will be able to fine tune gene editing therapies and RNA therapeutics.  
• Personalized cancer vaccines will become more common. Combined with a rise in AI models, RNA research is in a prime position to rapidly advance gene editing technologies to develop personalized cancer treatments. 
• More scientists will study the “dark genome,” or sections of DNA that do not code for traditional proteins. Preliminary research suggests that small proteins, such as micropeptides, in these DNA sections impact health more than previously thought. Assays that can code for these molecules, such as eRibo Pro, will be vital for exploring the dark genome. 

From everyone at Eclipsebio, thank you to our partners and the larger RNA community for an incredible 2025. We look forward to supporting you in 2026. 

To every member of our team who has given their energy, and dedicated their expertise, talent, and creativity into our mission – thank you. You are the reason Eclipsebio continues to blaze forward, helping our partners succeed in advancing RNA therapeutics.

Reflecting on our progress

In 2024, we launched eMERGE, our end-to-end sequencing-based platform for RNA therapeutics characterization. Industry adoption, by biotech innovators and major pharma partners alike has been amazing. It is a testament to the growing need for our advanced sequencing-based analytics that we are now supporting many of the 30 global biopharma advance their mRNA, saRNA and circRNA drug candidates to the clinic with key insights into RNA structure, translation, and purity.

While eMERGE anchors our offering, we have also expanded our portfolio in several key dimensions:

• Our miR-eCLIP+ assay addresses a major challenge in siRNA drug development: the direct measurement of off-target activity. This year, partners have rapidly adopted this technology to optimize and validate the next generation of RNAi therapeutics
• Our eVERSE database continues to gain traction as a foundation for AI model training in RNA therapeutics, helping close the data gap that has limited the effectiveness of AI in target discovery and drug design.
• We have expanded our sequencing capabilities to include nanopore sequencing, giving partners the ability to directly sequence RNA and gain insights into RNA quality that are not possible with other technologies.

These advances reflect our continued commitment not simply to provide analytical support, but to partner deeply with those developing the next generation of RNA medicines. Our brand promise remains the same: data-driven insights, not predictions; clarity, not black boxes.

The growing needs of RNA therapeutic development

RNA therapeutics continue to move beyond promise into clinical reality. With major regulatory milestones achieved, new modalities emerging, and manufacturing standards maturing, the field is shifting from “Can we do it?” to “Can we do it well?”

At Eclipsebio, we believe advanced, sequencing-based analytics are no longer optional: they are the foundation of high-quality RNA therapeutic development.

Over the past year, we have seen our partners focus on three key trends:

• Quality first: As the field expands from infectious disease vaccines to new classes of therapies such as mRNA-based gene editing, our partners are asking what RNA quality looks like when there is a critical need to minimize off-target effects and immune activation.
• Data scale: As cell-type-specific drug activity becomes increasingly important, multidimensional datasets are providing a competitive edge for those training effective AI models.
• Manufacturing analytics: The industry is reaching the limits of what traditional analytics can offer for guiding Quality by Design. There is growing demand for advanced sequencing-based analytics that illuminate the black box of RNA quality.

These are not incremental changes; they are redefining how RNA companies approach development, manufacturing, and commercialization.

Celebrating our culture and community

I am continually inspired by how our team shows up each day: solving complex problems, collaborating across disciplines, and staying relentlessly focused on partner success.

This year we were thrilled to support several early-stage drug companies and academic researchers through our Data Catalyst Award and RNA Innovators Award to provide rich RNA data for the design and characterization of tomorrow’s RNA medicines.

We also welcomed 5 new team members and celebrated key product launches, computational and AI advances, and revenue milestones including:

• Generating hundreds of thousands of data points from our advanced NGS assays to support our partners as they develop AI models for RNAi target discovery.
• The development of our own cutting-edge AI models for predicting the dynamics of translation, supporting the design of next-generation RNA therapies.
• Even deeper characterization of IVT RNA drugs with direct RNA sequencing.

To our partners, thank you for trusting us with your work. To our team, thank you for your passion, your insights, and your integrity. And to the next generation of scientists, engineers, and innovators, the opportunities to change the world with ground-breaking new RNA-based medicines have never been more immense. We all play a critical role and should embrace the challenges and journey together.  

Looking ahead

As I reflect on 2025, I feel a deep sense of optimism. The RNA field continues to unlock therapeutic pathways once considered out of reach. Eclipsebio is uniquely positioned to enable this progress through innovation, partnership, and a relentless focus on solving our partners’ greatest challenges.

Here’s to the breakthroughs yet to come and to the team that will make them possible. Together, let’s continue to turn RNA promise into patient impact.

Thank you for being part of this journey.

Progress in RNA therapeutics is increasingly supported by AI models trained to recognize the molecular patterns that define RNA behavior, such as how it folds, translates, or interacts with proteins. Yet, even the most sophisticated AI model depends entirely on the foundation it learns from: the data.

In RNA therapeutics, the same problem keeps surfacing. Many RNA datasets were never built for AI. They’re incomplete, inconsistently generated, or too shallow to capture the biological complexity needed for accurate predictions. As a result, models trained on them often fall short, performing well in testing but producing unpredictable results once applied in practice.

Truly “AI-ready” RNA data demands a higher standard. They are reproducible, multidimensional, and capture the biological context that drives functional outcomes. Together, these three attributes allow raw sequencing data to become machine learning models that can actually learn from and that drug developers can rely on.

1. Reproducibility

Every AI model depends on trust in its inputs. When replicate experiments don’t agree, a model learns technical noise instead of biology. True reproducibility doesn’t mean identical results; it means consistency when experiments are performed under the same conditions. Having multiple replicates is essential to capture genuine biological variability while minimizing technical noise. Reproducibility is what enables AI to learn biology rather than noise.

The challenge is that reproducibility across RNA datasets has historically been inconsistent, as public repositories aggregate data from different labs, protocols, and sequencing depths. Metadata can be incomplete, and those technical differences are easily mistaken for biological effects. As a result, models then end up learning artifacts introduced during sample preparation or analysis rather than true biological relationships. 

For AI, those inconsistencies lead to unstable models. Small batch effects can outweigh real biological signal, causing performance to collapse when data from a new experiment or cell type are introduced.

AI-ready RNA data minimizes these issues through standardized protocols and transparent quality control metrics. Consistent experiments limit technical variation, while a greater number of biological replicates increases confidence that observed differences reflect biology rather than technical bias. Together, these factors enable models to recognize patterns that hold true across systems and experimental contexts.

For RNA-based drug development, this reliability matters at every stage. Predictive models for RNA folding, stability, or translation efficiency are only as strong as their replicates allow. Reproducibility is what transforms experimental results into knowledge that AI can build upon.

2. Multidimensional coverage of RNA biology

Reproducibility ensures models can learn accurately, but biological breadth determines how much they can capture. In RNA biology, truly AI-ready datasets combine multiple complementary assays, each measuring a distinct layer of RNA behavior, to provide a comprehensive view of how RNA functions within the cell. 

Traditional datasets often measure only one or two aspects of RNA, typically expression levels and sequence variation, but leave out the rest. Yet RNA function emerges from a network of structural, regulatory, and translational processes. When data captures only a single layer, models can’t uncover the relationships that drive biological outcomes.

AI-ready datasets bring these layers together, measuring not only the abundance of the RNA but also how each molecule folds, interacts, and translates within the cell. Together, these complementary measurements reveal the full landscape of RNA behavior and may include: 

• Structural accessibility to identify where proteins or small molecules can bind.
• UTR and miRNA interactions to model post-transcriptional regulation.
• Translation and ribosome occupancy to connect sequence design to protein yield.
• RNA modifications, such as m6A, that influence stability, localization, or immunogenicity.

When all these features are measured in parallel and unified into a single dataset, models can integrate them into a comprehensive understanding of RNA behavior, the type of biological context required to design more stable, potent, and predictable therapeutics.

Comprehensive, multidimensional data doesn’t just improve model performance; it expands the questions AI can answer. Instead of asking “Does this RNA express?”, researchers can ask “Why does this one express better than the rest?” or “Which combination of structure, modification, and binding makes this variant more stable?” That’s the shift from descriptive to predictive RNA biology, where integrated data give AI the context to connect molecular mechanisms with meaningful outcomes.

3. Clear links to functional outcomes

Beyond capturing the layers of RNA biology, truly AI-ready datasets must also connect those molecular features to measurable function.  In drug development, the ultimate test of the data’s value lies in how well they link properties such as folding, binding, or translation to outcomes like potency, stability, or safety.

Many datasets stop short of this link. They provide detailed molecular profiles but with limited functional context, making it difficult to relate RNA features to therapeutic performance. For AI, that’s a dead end. Without a defined functional readout, even the best-designed models can’t generate meaningful insights.

AI-ready RNA datasets close this gap by pairing molecular measurements with clear functional outcomes, such as whether a transcript translates efficiently, remains stable over time, or triggers unwanted immune activation. These outcomes anchor model training, turning correlation into mechanisms AI can learn from and predict.

Once those functional links are established, AI can do what it does best: generalize. A model trained on thousands of structure-function examples can predict how a new construct will behave before it’s ever synthesized. In turn, these predictions can accelerate sequence optimization, improve manufacturability, and reduce the need for experimental iterations.

For drug developers navigating tight timelines and regulatory milestones, that level of foresight can make the difference between promising data and a viable therapeutic.

Why AI-ready RNA data matters now

AI has moved from an experimental tool to an operational one, advancing the stage of RNA drug discovery from identifying new targets to optimizing manufacturing. Yet even the most advanced algorithms can only perform as well as the data behind them.

Building models that perform reliably across different cell types, constructs, or modalities requires datasets that are explicitly designed for AI. That means data that are reproducible, multidimensional, and complete, linking molecular measurements to biological function. Together, these qualities enable algorithms to learn the actual rules governing RNA behavior and the patterns that explain how an RNA performs, not just how it appears in sequence.

Public datasets will continue to be valuable for exploratory analysis, but purpose-built RNA data resources are becoming the foundation of serious AI-driven drug development. They enable teams to transition from proof-of-concept modeling to actionable platforms that inform design decisions, expedite experimental cycles, and foster regulatory confidence.

A path forward

The gap between what AI can do and what it’s achieving in RNA therapeutics comes down to data readiness. The field has the tools; what it needs now are datasets that match the biological complexity of RNA itself.

Generating that kind of data requires consistency, depth, multidimensional coverage, and clear functional context, qualities often missing from public repositories. With eVERSE, Eclipsebio provides comprehensive, AI-ready datasets for RNA target discovery and drug design, uniting key layers of RNA biology from structure and regulation to translation.

Is your team looking for high-quality RNA data to train or validate your AI models? Let’s talk. 

However, the analytical methods used to assess RNA quality haven’t evolved as quickly as the science. Traditional characterization assays, once sufficient for simpler molecules, can no longer keep up with the complexity of today’s RNA modalities. As regulatory expectations increase, relying on legacy techniques risks leaving critical quality attributes unseen, unresolved, and unoptimized.

Developers need data that reflect the full molecular reality and capture how identity, integrity, purity, stability, potency, and safety interconnect. Sequencing-based characterization provides that level of visibility, giving teams the assurance needed to support both optimization and regulatory readiness.

The limits of traditional RNA characterization

Standard analytical tools like gel electrophoresis, HPLC, and mass spectrometry have been the workhorses of nucleic acid profiling for decades.

While these methods deliver useful snapshots, such as verifying approximate size, assessing purity, or confirming the presence of full-length transcripts, they often fall short when developers need deeper insight.

Surface-level data can’t capture molecular complexity. Traditional assays reveal what’s visible: overall size, yield, or bulk purity. What they miss are the subtle but significant molecular events that can make or break therapeutic efficacy, like sequence misincorporations, structural misfolding, or fragmentation. These defects may be invisible to conventional QC but can lead to truncated proteins, poor translation efficiency, or immune activation in vivo.

They’re not built for emerging RNA modalities. With mRNA, saRNA, and circRNA now in the pipeline, the structural and functional diversity of RNA therapeutics has exploded. Circular and self-amplifying constructs bring new challenges in structure, replication dynamics, and intracellular behavior that require the establishment of new, empirically-driven quality control thresholds.

They don’t meet today’s regulatory demands. Agencies are increasingly asking developers to demonstrate a comprehensive understanding of critical quality attributes (CQAs): identity, integrity, purity, stability, potency, and safety. Legacy assays can only partially address these factors. As a result, developers face blind spots that can slow filings, require rework, or trigger follow-up requests during regulatory review.

Outdated methods create data gaps and obscure the insights that can guide process optimization and ensure consistent performance from early development through manufacturing.

How sequencing-based characterization delivers a complete view of RNA quality

Sequencing-based technologies have opened new possibilities for understanding RNA therapeutics at single-nucleotide resolution. Rather than inferring quality from indirect measurements, these approaches observe it directly and capture the molecular fingerprints that define how an RNA behaves, performs, and persists.

Here’s what sequencing-based characterization can reveal that traditional assays cannot:

Identity: Confirm the right RNA, every time

Regulators expect the drug substance to match its intended design in nucleotide sequence and secondary structure. Sequencing provides direct confirmation that every base is correct and that folding patterns remain consistent across manufacturing lots. Developers can also detect rare misincorporations or structural deviations early and prevent the production of incorrect or misfolded proteins that compromise efficacy.

Integrity: Detect fragmentation before it impacts translation

RNA’s inherent instability means breakage can occur during transcription, purification, or storage. While commonly used assays show gross degradation, they can’t pinpoint where breaks occur or why. Sequencing maps fragmentation sites to reveal patterns linked to secondary structure or process conditions. This data allows developers to identify vulnerable regions and improve RNA design or manufacturing steps before they affect translation.

Purity: Map impurities at their source

Double-stranded RNA (dsRNA) impurities are among the most concerning for developers because they can activate innate immune responses and reduce therapeutic potency. Traditional antibody-based assays measure total dsRNA levels but can’t specify which regions of the RNA form these structures or why. Conversely, sequencing-based methods can directly identify dsRNA species and trace them back to their sequence origins—offering a roadmap for reducing impurity generation through data-informed design changes.

Stability: Predict degradation and performance

Stability defines how well an RNA therapeutic endures. Sequencing reveals which bases are most prone to hydrolysis maps their susceptibility under different conditions or formulations. It also connects structure to stability to clarify how modifications, buffers, or LNP encapsulation can affect RNA longevity.

Potency: See beyond expression levels

Traditional potency assays often measure protein output as a proxy for success, but they don’t capture why some constructs outperform others. Sequencing-based profiling tracks the complete translational journey from cellular uptake and endosomal escape to ribosome engagement and pausing. Gleaning such insights enables developers to refine sequence design, codon optimization, and delivery strategies for stronger, more predictable protein production.

Safety: Understand cellular responses in full

Even when an RNA product appears functional, off-target effects can arise at the transcriptional or translational level. Sequencing provides a comprehensive view of how cells respond after exposure, identifying gene expression or translation changes that may indicate unwanted immune or stress responses. Detecting these early supports safer candidate selection before clinical studies.

Building regulatory confidence through comprehensive data

Sequencing-based RNA characterization strengthens the evidence base that teams need to demonstrate control and consistency in regulatory submissions.

Alignment with CQA expectations

Regulatory agencies increasingly emphasize detailed characterization of the drug substance and drug product. Sequencing-based methods provide the robust datasets needed to demonstrate control over each CQA (identity, integrity, purity, stability, potency, and safety).

Consistency and traceability

Developers can compare lot-to-lot consistency at the molecular level, which builds confidence in reproducibility and control. Such a level of traceability supports comparability studies during process optimization and scale-up.

Risk mitigation and process optimization

Sequencing insights help teams detect issues early and correct design or manufacturing problems before they become regulatory roadblocks. This accelerates development timelines while reducing the likelihood of rework or additional testing during review.

A foundation for continuous improvement

Comprehensive RNA data not only satisfy regulators but also drive smarter development. Insights into structure-function relationships empower R&D and manufacturing teams to optimize performance and stability throughout the therapeutic’s lifecycle, enabling robust quality by design.

A new standard for RNA therapeutic characterization

As RNA medicines advance, a complete picture of molecular quality is no longer optional. Legacy assays can show what an RNA looks like, but can’t always explain how it behaves. Comprehensive characterization platforms like eMERGE use sequencing-based analyses to show how potency, stability, and safety influence therapeutic performance. The results inform decisions at every stage of development and manufacturing.

Whether you’re preparing for regulatory submission or refining your approach to RNA characterization, contact us to learn how our solutions can help.

• Traditional RNA sequencing methods rely on short reads, which can obscure or lose important information.
• Direct RNA sequencing with Oxford Nanopore preserves full-length RNA molecules and their base modifications.
• This approach reveals key insights such as modification locations, isoform identification, and poly(A) tail length profiling.
• Direct RNA sequencing is also a powerful tool for RNA therapeutic quality control, enabling confirmation of sequence identity, detection of byproducts, and monitoring of modifications.

Introduction

Next-generation sequencing (NGS) technology has unlocked countless insights into the fundamentals of biology and made tangible improvements to human health. However, when it comes to understanding RNA, most NGS techniques rely on indirect sequencing of RNA molecules. Typically, RNA is fragmented, converted into DNA, amplified, and then sequenced as short reads. This approach has many advantages, including scalability and cost-effectiveness, but these steps can also introduce biases and obscure information contained in the original RNA molecules.

An introduction to direct RNA sequencing

Platforms such as Oxford Nanopore preserve RNA features that short-read sequencing loses by offering the ability to sequence RNA molecules directly. This technology works by threading full-length RNA through a nanopore. As nucleotides pass through the pore, changes in electrical current are recorded. These fluctuations are then decoded using basecalling algorithms to determine the sequence of the original RNA molecule as it passes through the pore.

Unlike short-read methods, this technique does not require steps such as fragmentation or reverse transcription. As a result, it maintains critical layers of information about the starting RNA molecules. Sequencing and basecalling on Oxford Nanopore’s platform enable single-nucleotide resolution of base modifications, poly(A) tail length determination, and isoform identification. Simply put, the power of this approach comes from the ability to take intact RNAs and profile them from their 5’ start to their 3’ end without first manipulating or altering them.

Direct RNA sequencing in basic research

When profiling the transcriptome, these advantages allow researchers to explore aspects of RNA biology that are difficult to uncover with short-read sequencing. For example, a recent study used direct RNA sequencing to analyze how the transcriptome changed in response to knockdown of the m6A methyltransferase METTL3 in human cells¹.

Researchers profiled and characterized features such as RNA abundance, poly(A) tail length, the impact of alternative splicing on isoform identity, and the presence of base modifications across transcripts. This approach revealed new insights into RNA biology, including how METTL3 depletion influences RNA regulation and the interplay between poly(A) tail length and RNA stability.

Direct RNA sequencing for therapeutic QC

The utility of direct RNA sequencing extends beyond transcriptome studies. It is also a powerful tool for quality control (QC) of in vitro–transcribed RNA molecules intended for therapeutic use. We have previously discussed the importance of therapeutic RNA characterization and the features critical for designing successful mRNA medicines. Proper characterization is a multi-step process, beginning with informed design, continuing through evaluation of synthesis, and extending to efficacy testing and off-target profiling.

Direct RNA sequencing is especially relevant for QC after the production phase of this process. Developers of RNA medicines must confirm they are producing high-quality RNA, and long-read sequencing enables multiple assessments in a single experiment:

• Sequence identity: Detecting mutations such as substitutions, insertions, deletions, or rearrangements.
• Poly(A) tails: Measuring length and distribution, which affect stability and translation efficiency.
• Molecule integrity: Determining whether RNA is full-length or truncated.
• Base modifications: Confirming incorporation of modified nucleotides such as pseudouridine, which influence stability and immune activation.
• Byproduct detection: Identifying unintended transcriptional byproducts.

Importantly, a single round of direct RNA sequencing via Oxford Nanopore can provide information on all of these features simultaneously. This reduces processing steps, improves QC efficiency, supports consistency, and enables reliable monitoring of batch-to-batch variability. For example, Gunter et al. recently demonstrated how nanopore long-read sequencing (VAX-seq) can be used to simultaneously assess sequence identity, integrity, length, purity, and modification status in mRNA vaccine constructs². These considerations are critical for RNA medicines, where minimizing heterogeneity ensures reproducible dosing and patient safety.

Conclusion

Although direct RNA sequencing is powerful, it remains limited in certain aspects, such as throughput and accuracy, which make it a complement rather than a replacement for short-read technology. As the technology evolves, however, the insights it provides will only become more valuable.

Currently, RNA sequencing on the Oxford Nanopore platform offers an exciting way to examine RNA in its native state. By preserving full-length molecules, base modifications, and poly(A) tails, it provides insights into RNA biology that short-read methods alone cannot capture. For RNA therapeutics, from mRNA vaccines to emerging modalities, this approach adds a critical layer of quality control, ensuring that products are intact, properly transcribed, accurately modified, and consistent across production runs.

Such rigorous QC can bolster regulatory standards and help ensure that therapeutic RNAs consistently meet the highest quality benchmarks. As the field of RNA medicine continues to expand, integrating direct RNA sequencing alongside other advanced profiling methods will be key to accelerating discovery, improving design, and delivering safe, effective therapies to patients.

Whether you are advancing basic research or developing RNA therapeutics, our direct RNA sequencing services can help. Reach out to our team to learn more.

References

1. 1. Kim et al.
2. 2. Gunter et al.