The system, called NucleicBERT, applies a transformer-based language model to ribonucleic acid sequences, training it to predict masked positions in nucleotide strings drawn from large public databases. The approach deliberately avoids labels. Instead of being told which sequences are ribozymes, which are microRNAs, or which bind particular proteins, the model is simply asked to fill in the blanks across millions of natural sequences. In doing so, it is forced to internalise the statistical regularities of real RNA: which nucleotides tend to co-occur, which motifs recur across distant branches of life, and which combinations essentially never appear. Those patterns, the authors contend, encode a compressed representation of the physical and evolutionary constraints that shape functional RNA.

The technical foundation is the bidirectional encoder architecture popularised by models such as BERT. In natural language, such models read text in both directions and learn contextual embeddings, so that the meaning of a word depends on its neighbours. NucleicBERT imports that idea wholesale into molecular biology. Each nucleotide in an RNA sequence is treated as a token, and the encoder produces a vector for every position that reflects its biological context. A cytosine embedded in a stem-loop of a transfer RNA acquires a different representation from the same cytosine sitting in the loop of a riboswitch, even though the raw letter is identical. This context sensitivity is precisely what hand-crafted features, position-weight matrices and simple motif searches have historically lacked.

Pretraining proceeds with a masked-language objective. Random positions in each training sequence are hidden, and the model must reconstruct them from surrounding context. Because the training corpus spans diverse RNA families and organisms, the network cannot succeed by memorising shallow patterns; it must capture deeper regularities such as compensatory mutations in paired regions, conserved loops, and the compositional biases of different RNA classes. The resulting embeddings can then be transferred downstream: a relatively small amount of labelled data is sufficient to fine-tune the pretrained network for specific prediction tasks, a strategy that has transformed fields from computer vision to protein biochemistry.

The practical payoff comes in the form of benchmark performance on tasks that matter to RNA biologists. According to the paper, NucleicBERT embeddings improve predictive accuracy on problems including the classification of non-coding RNA families, the identification of RNA-binding protein sites, and the assessment of sequence variants that disrupt splicing or translation. In each case the pretrained model outperforms baselines trained from scratch on the same labelled data, and the advantage is largest precisely where labelled examples are scarcest. That pattern is the classic signature of useful pretraining: the model arrives at a task already fluent in the underlying vocabulary, so supervision only needs to teach the final grammar.

What makes the work conceptually significant is not merely the benchmark numbers but the interpretability experiments layered on top of them. The authors probe what the model has learned by examining attention patterns and embedding geometry. Sequences with related structures and functions cluster together in the embedding space even when their nucleotide identities differ substantially, suggesting the model has discovered homology that raw sequence comparison misses. Attention heads, the internal components that let a transformer weigh relationships between positions, turn out to concentrate on regions that biologists recognise as structurally or functionally meaningful, such as paired stems and conserved catalytic motifs. In effect, the network rediscovers, from raw data alone, some of the hard-won knowledge that RNA biochemists assembled over half a century.

The study also confronts one of the central puzzles of RNA biology: the sheer size of sequence space. An RNA molecule of only 100 nucleotides has 4 to the power of 100 possible sequences, a number that dwarfs the number of atoms in the observable universe. Natural RNA occupies a vanishingly sparse subset of that space, organised into families shaped by common ancestry and common physics. Language models are, in a formal sense, tools for modelling the distribution of data, and NucleicBERT can therefore be read as a statistical map of where functional RNA lives within the vast combinatorial wilderness. Sequences the model assigns high likelihood are, heuristically, sequences that look like biology; sequences it assigns low likelihood are candidates for exotic synthetic designs, or for failure.

That map has immediate applications in engineering. RNA therapeutics, from messenger RNA vaccines to small interfering RNAs and antisense oligonucleotides, all depend on the properties of sequence: how stably a molecule folds, how efficiently it is translated, how recognisable it is to the innate immune system, and how long it survives in the cell. The authors report that NucleicBERT representations correlate with measurable properties such as secondary-structure stability and expression level, offering drug developers a way to screen and optimise candidate sequences in silico before expensive synthesis and testing begin. The same representations can guide the design of synthetic riboswitches and regulatory elements for synthetic biology, where designers currently iterate through costly build-and-test cycles.

The researchers are candid about limitations. RNA databases are biased towards well-studied model organisms and abundant RNA classes, so the model’s fluency is strongest where data are richest and weakest for rare transcripts and poorly characterised clades. The masked-language objective captures linear sequence context directly and higher-order structure only indirectly, so tasks that hinge on detailed three-dimensional folding may still require complementary physics-based or structure-specific models. And like all deep networks, NucleicBERT offers correlations rather than mechanisms: its embeddings are a powerful substrate for prediction, but turning them into causal explanations of why a particular fold catalyses a particular reaction remains future work. The authors frame the model not as a replacement for biochemical experiment but as a hypothesis engine that tells experimentalists where to look.

Even with those caveats, the arrival of a mature nucleic-acid language model marks a turning point in how the life sciences approach sequence data. For twenty years, genome annotation has leaned on alignment-based tools that compare new sequences against known ones, a strategy that fails for molecules with no recognisable relatives. Self-supervised models offer a different epistemology: knowledge distilled from the totality of observed sequences, applicable even to orphans with no evolutionary cousins. As sequencing technologies continue to generate data far faster than any human can annotate them, systems like NucleicBERT are likely to become standard equipment in the computational biology toolkit, reading the genome’s least understood language at a pace no human reader could match and pointing the way to RNA molecules that biology has not yet invented.

Subject of Research: Self-supervised language modelling of RNA sequence space with the NucleicBERT neural network

Article Title: NucleicBERT interprets RNA sequence space through self-supervised language modelling

Article References: Upadhyay, U., Herold, J., Götz, M., & Schug, A. (2026). NucleicBERT interprets RNA sequence space through self-supervised language modelling. Nature Machine Intelligence. https://doi.org/10.1038/s42256-026-01295-9

Image Credits: AI Generated

DOI: 10.1038/s42256-026-01295-9

Keywords: NucleicBERT, RNA, self-supervised learning, language models, machine learning, transformers, non-coding RNA, RNA therapeutics, sequence biology, computational biology, embeddings, RNA structure

Traditional gene-level analysis methods have long dominated the study of cellular diseases, yet they are limited by their inability to capture the intricate variability present within each gene. Typically, these methods compress all the RNA data of a gene into a single count, effectively masking the subtleties and nuances that could better inform disease presence, progression, and treatment response. Recognizing this critical gap, the McGill team sought to transcend the constraints of conventional gene-level assessments by developing an approach that interrogates the smaller building blocks of genes—exons and their junctions.

DOLPHIN leverages advanced machine learning algorithms to analyze how the sections of genes, known as exons, are spliced and connected within RNA sequences of single cells. Unlike earlier methods that view genes as monolithic blocks, this tool embraces the modular nature of genetic material, akin to assembling LEGO bricks in various configurations. This exon- and junction-centric perspective unveils a previously uncharted level of cellular complexity and heterogeneity, which is invaluable for pinpointing disease markers that escape detection by standard techniques.

The scientific team demonstrated the tool’s exceptional capabilities through a compelling application on pancreatic cancer data. Pancreatic cancer is notorious for its aggressive nature and poor prognosis, often due to late detection and limited treatment options. DOLPHIN analyzed RNA sequencing data from individual cells within tumor samples and successfully uncovered over 800 disease markers that had eluded conventional gene-level analyses. More impressively, the AI tool could distinguish between patients harboring high-risk aggressive tumors and those with less severe disease forms, thus offering critical insights that could tailor therapeutic decisions and improve clinical outcomes.

Beyond the immediate diagnostic improvements, DOLPHIN’s contributions to the field of single-cell transcriptomics mark a pivotal step toward the ambitious goal of building comprehensive digital models of human cells. These “virtual cells” hold the promise of simulating cellular behavior and predicting responses to pharmaceutical compounds in silico, significantly reducing the need for labor-intensive and costly laboratory or clinical trials. By generating richer and more precise single-cell profiles, DOLPHIN lays the groundwork for these transformative digital simulations, which could redefine the future of biomedical research and drug development.

The research team acknowledges that while the initial results are encouraging, scaling the tool to analyze millions of cells across diverse datasets is an essential next phase. Such expansion will enhance the resolution and accuracy of virtual cell models, helping to capture the full spectrum of cellular states and disease manifestations across different tissues and patient populations. This scalability will be vital for integrating DOLPHIN into routine biomedical workflows and translating its benefits from the laboratory to the clinic.

Central to the tool’s success is its ability to exploit the vast amount of information contained within exon and junction reads—elements often overlooked by traditional analyses. These reads represent the transcriptomic intricacies of how genes are pieced together post-transcriptionally, influencing cell function and identity. By effectively interpreting this layered information, DOLPHIN transcends simplistic gene expression counts and embraces the dynamic nature of gene regulation, which is frequently altered in diseases such as cancer.

Furthermore, DOLPHIN’s AI-driven methodology blends computational prowess with biological insight, exemplifying the interdisciplinary synergy necessary to tackle complex health challenges. The model’s capability to process and learn from massive and multidimensional datasets sets a precedent for future tools aiming to decrypt cellular behavior with comparable depth and precision. Its application thus reflects the transformative potential of computational biology in ushering in an era of precision medicine.

The broader implications of DOLPHIN extend beyond oncology. The ability to detect subtle RNA splicing alterations and disease markers at the single-cell level might illuminate the molecular underpinnings of a wide array of conditions, from autoimmune disorders to neurodegenerative diseases. Such advances could enable earlier detection, more accurate prognostication, and personalized treatment plans that vastly improve patient quality of life.

This research, spearheaded by Kailu Song, a PhD student in McGill’s Quantitative Life Sciences program, along with senior author Jun Ding, an assistant professor in the Department of Medicine and a junior scientist at the Research Institute of the McGill University Health Centre, represents a compelling leap forward in single-cell analysis. Their study, published in the renowned journal Nature Communications, underscores the power of refining transcriptomic data to unlock hidden cellular information vital for medical innovation.

Funded by prestigious organizations such as the Canadian Institutes of Health Research, the Natural Sciences and Engineering Research Council of Canada, and the Fonds de recherche du Québec, this project exemplifies the critical role of sustained research investment in driving cutting-edge scientific discovery. The confluence of AI technology with molecular biology heralded by DOLPHIN is a testament to how collaborative, interdisciplinary efforts can reshape the future of health care.

As DOLPHIN continues to evolve and integrate into diverse biomedical investigations, its promise to chart uncharted territories within the cellular genome remains unparalleled. By unveiling the finer details of genetic regulation hidden within single cells, this AI tool not only enhances our understanding of disease mechanisms but also sparks new hope for earlier diagnosis, more effective treatment, and ultimately, better patient lives worldwide.

Subject of Research: Cells
Article Title: DOLPHIN advances single-cell transcriptomics beyond gene level by leveraging exon and junction reads
News Publication Date: 4-Jul-2025
Web References: https://www.nature.com/articles/s41467-025-61580-w
References: Song, K., Ding, J., et al. (2025). DOLPHIN advances single-cell transcriptomics beyond gene level by leveraging exon and junction reads. Nature Communications. DOI: 10.1038/s41467-025-61580-w
Keywords: Cell biology, single-cell transcriptomics, artificial intelligence, exon splicing, pancreatic cancer, precision medicine, RNA sequencing, computational biology