The challenge that motivated the work is rooted in the physics of nanopore sequencing. As an RNA or DNA molecule threads through a protein pore, the sequencer records characteristic disruptions in ionic current, and basecalling algorithms translate those disruptions into sequence. Homopolymer runs, long stretches of a single base such as the adenines in a poly(A) tail, produce nearly identical current signals base after base, making it hard to count exactly how many adenines are present. On top of that, the 3′ ends of transcripts are architecturally complex: the boundary between the end of the transcript proper and the start of the tail can be fuzzy, alternative polyadenylation sites can generate multiple transcript isoforms from a single gene, and repetitive elements, including remnants of endogenous retroviruses, can complicate the identification of where a transcript truly terminates.

PolyAnalysis tackles these problems at the signal level rather than working from basecalled sequence alone. The framework combines a convolutional neural network with a bidirectional long short-term memory network, a pairing commonly abbreviated as CNN–BiLSTM, to classify regions of the raw signal as poly(A)-proximal or otherwise. It then applies a multi-task connectionist temporal classification decoding scheme, adapted from speech recognition, to infer the boundaries of the poly(A)-proximal region and the tail length itself. Crucially, the authors built adenine-aware training and decoding into the pipeline: the loss function used during training weights adenine-related errors differently, and the decoding step is biased toward A-rich interpretations, reflecting the biological expectation that the region of interest is dominated by adenines.

Benchmarking was carried out against three established tools: Nanopolish, tailfindr, and Dorado. The evaluation used controlled datasets generated with both RNA002 and RNA004 nanopore chemistries, allowing the team to assess performance across sequencing generations. According to the study, PolyAnalysis achieved tail-length accuracy that was competitive with the existing methods while delivering high callability, meaning it produced usable estimates for a larger fraction of reads. The authors stress that tail-length estimation is the most extensively validated output of the framework, and they are careful not to overstate the maturity of the other analyses the tool offers.

To understand which components of the model actually mattered, the researchers performed component-wise ablation experiments, systematically removing or altering parts of the pipeline and measuring the effect on performance. They also ran bias-control analyses. The results indicated that the region classification module, the multi-task optimization objective, the adenine-weighted loss, and the A-rich decoding strategy each contributed to the overall performance, suggesting that the design choices were not arbitrary but jointly responsible for the accuracy gains. This kind of ablation work is increasingly expected in machine learning applied to genomics, where it is easy to attribute improvements to a single flashy component when in fact the benefit comes from an interplay of design decisions.

Beyond tail length, PolyAnalysis links its per-read estimates to analyses of polyadenylation sites, alternative polyadenylation, and repetitive-element-associated 3′ ends. When applied to human cell-line datasets and vertebrate transcriptome datasets, the framework identified variation in tail-length distributions at the dataset level and heterogeneity at the gene level. It also found only weak associations between tail length and steady-state transcript abundance, a finding that adds nuance to the widely held expectation that longer tails should reliably predict more abundant or more translated transcripts. The relationship between tail length and expression, the data suggest, is more context-dependent than simple models imply.

One of the more distinctive features of the framework is its confidence-based filtering of polyadenylation sites. Rather than presenting all detected sites as equally reliable, PolyAnalysis separates them into database-matched high-confidence sites, novel candidates flagged with high confidence, and low-confidence calls. This tiered output acknowledges a persistent problem in 3′ end profiling: novel polyadenylation sites predicted from sequencing data can reflect genuine biology or merely artifacts of alignment, basecalling, or annotation gaps. By making confidence explicit, the tool gives downstream users a principled way to decide which calls to trust for follow-up experiments.

The treatment of endogenous retroviruses illustrates the same caution. Retroviral remnants scattered through the genome can supply polyadenylation signals to nearby genes, and distinguishing true locus-level evidence of such events from family-level patterns, where many similar retroviral sequences produce ambiguous signals, is genuinely difficult. PolyAnalysis applies ambiguity-aware analysis to separate these two levels of evidence, and the authors are explicit that the sequence-composition, polyadenylation-site, alternative-polyadenylation, and endogenous-retrovirus-related results all remain dependent on model assumptions, sequencing chemistry, and annotation quality, and require further orthogonal validation before being treated as established biology.

The significance of the work lies partly in its scope. Alternative polyadenylation is a major layer of gene regulation: by choosing a proximal or distal polyadenylation site, a cell can change the untranslated region of a transcript and thereby alter its stability, localization, and translation efficiency, with documented roles in development, immunity, and cancer. Tools that can profile these choices per read, on long-read platforms that capture full transcript molecules, offer a view that short-read sequencing cannot easily provide. By coupling tail-length estimation with polyadenylation-site analysis in a single signal-level framework, PolyAnalysis aims to make that view more accessible to laboratories that already run nanopore sequencers.

The study, a software contribution from researchers at Hunan University, Hunan University of Finance and Economics, and the University of Electronic Science and Technology of China, was supported by the National Natural Science Foundation of China and involved secondary analysis of publicly available datasets. The framework is published open access under a Creative Commons Attribution license, and the authors have released supplementary figures, tables, and numerical source data underlying the main results. For a field where poly(A) tail measurement has often been a bespoke, error-prone exercise, a benchmarked, confidence-aware pipeline that works across both RNA002 and RNA004 chemistries represents a practical step forward, even as the authors themselves caution that the most exploratory parts of their analysis, particularly the retrovirus-associated findings, should be treated as hypotheses awaiting independent confirmation rather than settled results.

Subject of Research: Deep learning-based estimation of poly(A) tail length and 3′-end regulation profiling from nanopore sequencing data

Article Title: PolyAnalysis: a deep learning–based framework for nanopore-based poly(A) tail estimation and 3′-end regulation profiling

Article References: Tian, Q., Song, B., Zou, Q., & Wang, Y. (2026). PolyAnalysis: a deep learning–based framework for nanopore-based poly(A) tail estimation and 3′-end regulation profiling. BMC Biology. https://doi.org/10.1186/s12915-026-02749-7

Image Credits: AI Generated

DOI: 10.1186/s12915-026-02749-7

Keywords: poly(A) tail, nanopore sequencing, deep learning, alternative polyadenylation, polyadenylation site, long-read transcriptomics, connectionist temporal classification, endogenous retrovirus, RNA sequencing, 3′ end regulation, Oxford Nanopore Technologies, BMC Biology

The research, led by J. P. Silva and J. L. Fiedler at the Laboratory of Neuroplasticity and Neurogenetics of the Universidad de Chile, focused on the CA1 region of the dorsal hippocampus, a area essential for forming and retrieving spatial and episodic memories. Rather than treating the tissue as an undifferentiated soup of cells, the team physically separated two functionally distinct compartments: the soma, where neuronal cell bodies house the nucleus and most of the cell’s protein-making infrastructure, and the neuropil, the dense thicket of axons, dendrites and synapses that surrounds those cell bodies and carries out the actual business of neural communication. This spatial split matters because neurons are known to ship specific messenger RNAs out to remote dendritic compartments, where they can be translated into proteins on-site, a process considered essential for synaptic plasticity, the cellular substrate of learning and memory.

To read the RNA in each compartment, the researchers turned to long-read direct RNA sequencing from Oxford Nanopore, a technology that threads intact, full-length RNA molecules through protein nanopores and reads them without the fragmentation and assembly steps required by conventional short-read methods. Where standard RNA sequencing often struggles to distinguish which exon combinations belong to which transcript, the long-read approach resolves individual isoforms, the alternative versions of a gene’s message produced when different exons are stitched together, when transcription starts or stops at different points, or when the poly(A) tail at the message’s end is trimmed or extended. In this study, the technique allowed the team to identify full-length transcripts, quantify isoform usage, and even detect RNA species that had never been annotated in the rat genome’s reference catalog.

The first surprise was a kind of parity. When the researchers compared male rats with females in the estrus phase, a stage of the reproductive cycle characterized by relatively low estrogen levels, they found that the overall numbers of genes and isoforms detected in CA1 were broadly comparable between the sexes. Roughly eighty percent of the transcripts they detected were classified as protein-coding messenger RNAs, and about eight percent corresponded to RNA species absent from existing annotations, a reminder of how much of the transcriptome remains unmapped even in a tissue as intensively studied as the hippocampus. At the coarse level of counting molecules, male and female CA1 looked reassuringly similar.

The differences emerged when the researchers looked at where the molecules were and which versions of them were being used. Sex-associated differences proved more pronounced in the somatic compartment, where males showed a higher variety of RNA isoforms than estrus-phase females, while transcript profiles in the neuropil diverged far less between the sexes. Differential expression analysis told a complementary story: in estrus-phase females, 892 transcripts were enriched in the neuropil and 467 in the soma, whereas in males the corresponding figures were 973 neuropil-enriched and 541 soma-enriched transcripts. Intriguingly, some RNA instructions that were abundant in the neuropil of males were instead more concentrated in the cell bodies of females, suggesting that the two sexes route their molecular cargo to different addresses within the same neurons.

The study also uncovered structural signatures that distinguish the two compartments regardless of sex. Transcripts enriched in both the soma and the neuropil carried longer 5-prime untranslated regions, the stretches of RNA upstream of the protein-coding sequence that regulate how efficiently a message is translated. Somatic transcripts, however, bore longer 3-prime untranslated regions, segments that harbor binding sites for microRNAs and RNA-binding proteins and thus govern message stability and localization, while short poly(A) tails emerged as a hallmark of the neuropil fraction. Because poly(A) tail length can influence both translation efficiency and mRNA decay, this compartmental difference hints at distinct post-transcriptional control regimes operating in cell bodies versus synaptic territories.

One of the most striking sex-specific findings was that female neuropil-enriched transcripts displayed significantly shorter coding sequences than their male counterparts. Functional enrichment analysis added texture to this observation: transcripts enriched in the female neuropil were associated with lipid metabolism and translation-related processes, suggesting a neuropil environment tuned toward local protein synthesis and membrane-related functions. Male-enriched somatic transcripts, by contrast, were linked to memory, the regulation of cellular activity, and the development of neuronal projections, a functional profile consistent with the soma’s role as the command center from which neuronal architecture is built and maintained.

At the level of individual isoforms, the analysis revealed sex- and layer-specific enrichment of RNAs encoding receptors, scaffolding proteins, ribosomal components, and transcriptional regulators. This is a consequential category of molecules. Receptors determine how a neuron responds to its inputs, scaffolding proteins organize the molecular machinery of the synapse, ribosomal components set the capacity for local protein production, and transcriptional regulators feed back to reshape the cell’s entire gene expression program. Sex-dependent selection among isoforms of such genes means that male and female CA1 neurons may not merely express different amounts of the same proteins, but different versions of them, with potentially distinct interaction partners, localization signals, and regulatory properties.

The authors are careful to frame their findings as a baseline map rather than an explanation of behavioral differences. The comparisons were made under basal conditions, and the female cohort was restricted to the estrus phase, one specific hormonal state within the estrous cycle. Since estrogen and other ovarian hormones are known to modulate hippocampal gene expression and spine density, examining other phases of the cycle, or hormonally manipulated animals, would be a natural next step. Nevertheless, the study demonstrates that the molecular landscape of the hippocampus is organized along two axes simultaneously, compartment and sex, and that these axes interact in ways that shorter-read, whole-tissue approaches would have averaged away entirely.

The broader significance lies in the methodological shift the work exemplifies. For years, transcriptomic studies of the brain have catalogued which genes are switched on and by how much, but the layer of biology that determines which isoform of each gene is deployed, where in the neuron it travels, and how its untranslated regions and poly(A) tail are configured has remained largely invisible. By resolving that layer in a memory-critical brain region, and by showing that it differs between the sexes even at rest, the Chilean team has provided a molecular framework for investigating why male and female brains differ in their susceptibility to neurological and psychiatric disorders, from Alzheimer’s disease to depression, conditions in which both hippocampal dysfunction and sex bias are well documented. The map is drawn in rats, but the questions it opens apply to brains of every species, including our own.

Subject of Research: Sex-associated isoform distribution in the soma and neuropil of the rat hippocampal CA1 revealed by long-read direct RNA sequencing

Article Title: Long-read RNA sequencing reveals sex-associated isoform distribution in soma and neuropil of the rat hippocampal CA1

Article References: Silva, J. P., Catalán, J., Olave, F. A., González, P. I., Guarnieri, T., Palacios-Avendaño, N., Corrales, W. A., Alarcón-Mardones, M., Aliaga, E., Munita, R., Maracaja-Coutinho, V., & Fiedler, J. L. (2026). Long-read RNA sequencing reveals sex-associated isoform distribution in soma and neuropil of the rat hippocampal CA1. Biology of Sex Differences. https://doi.org/10.1186/s13293-026-00988-5

Image Credits: AI Generated

DOI: 10.1186/s13293-026-00988-5

Keywords: long-read sequencing, RNA isoforms, hippocampus, CA1, sex differences, neuropil, soma, transcriptome, Oxford Nanopore, poly(A) tail, synaptic plasticity, gene expression