Biology subjects

Lu, S.

Publications and source records attributed to Lu, S..

At least 19 recordsLinked to original sources

bioRxiv2026

Single-Nucleus Profiling Reveals Neuronal Metabolic Collapse During the Transition to Alzheimer's Disease Dementia

Current Alzheimer's disease (AD) neuropathological metrics, including Thal phasing, and Braak staging, often do not accurately predict cognitive status: many cognitively resilient individuals have substantial pathological burdens, whereas some patients with dementia show minimal classical AD pathology. Here, we addressed this discrepancy by investigating the molecular mechanisms underlying AD dementia using single-nucleus RNA-sequencing and single-nucleus multiome datasets. To eliminate cross-modality scaling artifacts and compositional biases inherent in neurodegenerative tissues, we normalized the datasets using our novel framework, bypassing scaling issues of conventional methods that may obscure critical biological signals of interest. We identified neuron-specific suppression of nuclear-encoded mitochondrial genes and their upstream mitochondrial biogenesis signaling (MBS) as primary drivers of clinical dementia. By integrating multi-omic quantitative trait loci, genome-wide association studies, and epigenomic profiles, we delineated genetic and epigenetic changes underlying MBS suppression. This study provides a systematic genomic framework for understanding the mechanisms of cognitive resilience.

bioinformatics↗

bioRxiv2026

Gap-free telomere-to-telomere haplotypes assembly of the hybrid grouper 'Yushuban' (Cromileptes altivelis female x Epinephelus fuscoguttatus male)

The hybrid grouper 'Yushuban' (Cromileptes altivelis [female] x Epinephelus fuscoguttatus [male]) has shown promising performance in aquaculture trials, yet no complete reference genome has been available for this hybrid. We generated two gapless, telomere-to-telomere (T2T) haplotype-resolved genome assemblies of a single individual by integrating short-read, PacBio HiFi, Oxford Nanopore ultralong-read and Hi-C data. The maternal (Yushu_Calti_1.0) and paternal (Yushu_Efusc_1.0) haplotypes are 1,077.33 Mb and 1,083.36 Mb in length, with N50 values of 45.46 Mb and 45.55 Mb, respectively. Across all platforms, over 99.4% of reads mapped to each assembly, and Merqury-based quality values reached 57.7413 and 59.2068 (PacBio). BUSCO recovered 99.78% and 99.75% of complete core genes. We predicted 27,248 and 27,566 protein-coding genes, of which 98.52% and 99.23% were functionally annotated, and annotated telomeric motifs and repetitive elements (46.12% and 47.18% of each genome). These assemblies fill a gap in grouper genomic resources and provide a foundation for genetic dissection, parental genome comparison and molecular breeding of this hybrid.

genomics↗

bioRxiv2026

MINT infers the latent single-cell spatial transcriptome from paired histology

Sequencing-based spatial transcriptomics captures transcriptome-wide molecular information, yet the corresponding cell-resolved tissue map remains incomplete. Histology records dense cellular morphology, tissue architecture and local context. Here, we present MINT (Morphological Integrative Network for Transcriptomics), which combines this latent cellular organisation with sample-specific spatial measurements to infer tissue-wide, cell-resolved transcriptomes without an external single-cell reference. MINT reconstructs expression from sparse Slide-tags profiles across unmatched adjacent sections; its shared nucleus-centred representation also resolves pooled Visium measurements into cell-level profiles. MINT improved Slide-tags alignment and expression recovery under controlled nuclear loss and cross-section perturbation. It also outperformed reference-free Visium disaggregation methods against Xenium ground truth and was validated with Visium HD sequencing. In human lung adenocarcinoma, MINT reconstructed the tumour immune microenvironment cell by cell and resolved tertiary lymphoid structures and their maturity. These results establish paired histology as a quantitative, sample-intrinsic complement to molecularly rich but cellularly incomplete spatial-transcriptomic assays.

bioinformatics↗

bioRxiv2026

Chromosome-Level Nuclear Genome of Simulium vittatum Links Black Fly Cytogenetics to Genome Organization and Evolution

Black flies (Diptera: Simuliidae) are important vectors of pathogens affecting human and animal health, yet the absence of a chromosome-level nuclear reference genome has constrained molecular and evolutionary studies of the family. Here, we present the first chromosome-level nuclear genome and a developmental transcriptomic resource for the long-established IS-7 laboratory lineage of Simulium vittatum. Combining Oxford Nanopore long-read sequencing with Hi-C scaffolding, we assembled a 340.4-Mb genome, with 99.1% of the assembly resolved into three chromosome-length scaffolds (N50=104.8 Mb, BUSCO completeness 93.9%), consistent with the known 2n = 6 karyotype. Using the historically mapped molecular landmarks SVAT and SVEP, we assigned the two arms of chromosome III as IIIS and IIIL, respectively, linking sequence coordinates to the classical polytene chromosome map. Repetitive DNA comprises 46.48% of the assembly, including 29.68% unclassified repeats, and annotation identified 14,732 protein-coding genes and 16,417 transcripts. Integration with larval RNA-seq recovered the previously characterized silk gland proteins SGP-1-SGP-3 and resolved the expressed SGP-2 sequence despite a remaining discrepancy with the genomic sequence. This reference genome connects classical black fly cytogenetics with sequence-level analyses of genome organization, structural variation, and gene content, addressing a major genomic gap within Culicomorpha and providing a foundation for comparative studies of chromosome evolution, hematophagy, and vector biology across Simuliidae.

genomics↗

bioRxiv2026

mBaoJin-labeled pangolin coronavirus for evaluating population cross-neutralizing antibodies and the entry-inhibitory activity of cepharanthine

Replication-competent coronaviruses carrying fluorescent protein-tagged structural proteins remain scarce. Using the highly attenuated pangolin coronavirus GX_P2V(short_3UTR) as a backbone, we generated GX_P2V-mBJ-N, a recombinant coronavirus in which the bright green fluorescent protein mBaoJin is fused to the nucleocapsid (N) protein. The reporter virus is attenuated and genetically unstable in normal Vero cells but can be amplified to high titers in cells expressing wild-type N, and its fluorescence directly reports N protein expression. Using this authentic-virus platform, we show that high-titer GX_P2V cross-neutralizing antibodies persist in most healthy individuals and that cepharanthine potently blocks viral entry. GX_P2V-mBJ-N thus provides a simple and reliable tool for coronavirus tracing, immune surveillance, and antiviral drug evaluation.

microbiology↗

bioRxiv2026

Virtual-cell verification enables self-auditing AI discovery for immune rejuvenation

Artificial-intelligence agents propose drug-discovery hypotheses faster than experiments can test them, yet their conclusions are rarely verified, against the underlying biology, the predicted perturbation, or the agent's own scoring logic. We close this verification gap with an agentic framework built on three verifiers. First, PACE, a phenotype verifier, resolves immune aging into ten directionally scored, cell-type-resolved gene-set modules, selected for cross-cohort stability across four PBMC cohorts, and outperforms five established aging clocks in an independent in-house aging cohort of 434 elderly donors. Second, CellQ, a virtual-cell verifier built with multi-modal LLM, compresses each single-cell transcriptome into eight discrete tokens aligned to a language model's vocabulary through residual vector quantization; it attains state-of-the-art perturbation prediction and uniquely resolves the weak, module-level shifts that differential-expression recovery misses. Third, an Analyzer-Planner-Auditor agent verifies its own scoring logic: screening 110 compounds in primary human PBMCs, it found aged-down modules more reversible than aged-up modules and revised its objective from an equal-weight mean to a balance-constrained minimum, a self-correction that generalized to an independent 13-compound T-cell assay. By verifying its predictions and its own objective against experiment, the framework points beyond hypothesis-generating AI toward self-correcting AI scientists whose objectives could continuously evolve.

bioinformatics↗

bioRxiv2026

Climate warming reduces the speed and predictability of polygenic adaptation to salinity decline

Climate change exposes populations to multiple stressors simultaneously, yet our understanding of how the addition of one stressor alters adaptation to another remains poor. As a result of climate change, high-latitude coastal habitats are experiencing rapid salinity decline, resulting in serious impacts on food webs and ocean circulation. Here, we examine how temperature increase impacts adaptation to salinity decline, in terms of the speed, genomic response, and repeatability of adaptation. We performed replicated Evolve-and-Resequence experiments over 20 to 25 generations in the model copepod Eurytemora carolleeae (Atlantic clade of the E. affinis species complex). Under salinity decline alone, replicate selection lines exhibited a polygenic response involving 66 selected haplotype blocks, with increasing parallelism among the replicate lines through Generation 20. Fitness (egg number) declined sharply over the first four generations but underwent full Evolutionary Rescue, recovering to ancestral levels by Generation 10. In contrast, imposing temperature increase on the salinity decline lines resulted in significantly lower parallelism among the selection lines, along with delayed and incomplete Evolutionary Rescue. Only 14% of selected SNPs were shared between the two selection regimes, and Gene Ontology analyses revealed largely distinct functional categories of genes under selection. These results show that adding warming to salinity decline can alter the genomic trajectory of salinity adaptation, slowing and impeding Evolutionary Rescue, and reducing the repeatability of polygenic responses. Our findings have direct implications for predicting evolutionary responses to realistic, multi-stressor climate change in high-latitude coastal ecosystems experiencing simultaneous ocean freshening and warming.

evolutionary biology↗

bioRxiv2026

Pharmacologic eIF2B Activation Rescues Neuropathy in CMT2 Subtypes by Normalizing the Integrated Stress Response

Among the many subtypes of Charcot-Marie-Tooth (CMT) disease, several result from mutations in genes encoding aminoacyl-tRNA synthetases, enzymes required for tRNA charging during cytoplasmic and mitochondrial translation. We report that activation of the integrated stress response (ISR) pathway is a shared molecular feature of tRNA synthetase-associated and other axonal CMT subtypes. RTX-117, a CNS-penetrant small molecule currently in Phase 1 clinical trials, targets eukaryotic initiation factor 2B (eIF2B), a key modulator of protein synthesis and the ISR pathway. Using cryo-EM studies, we have characterized the binding mode of RTX-117 to the eIF2B decamer. In GarsP278KY/+ mice, which develop early onset motor defects and axonal pathology that recapitulate CMT2D symptoms from tRNA synthetase mutations, RTX-117 treatment started after disease onset reduced chronic ISR activation and produced significant functional and electrophysiological improvement. We further identify ISR targets, including secreted proteins such as GDF15 and FGF21 that may serve as translational biomarkers for treatment response to RTX-117 in CMT disease. Broader surveillance of the ISR pathway across models of neurodegeneration reveals strong activation in several diseases and a correlation with disease progression, particularly in models of Alzheimers disease. These findings identify chronic ISR activation as a recurrent, though not universal, pathological mechanism of neurodegenerative disease models. Overall, our study identifies candidate biomarkers for CMT disease subtypes associated with defects in translational homeostasis and supports eIF2[a]-ATF4 axis modulation as a promising therapeutic strategy for this disease class. One Sentence SummaryRTX-117, a clinical stage eIF2B activator, blunts chronic ISR activation and improves nerve and motor function in a mouse model of Charcot-Marie-Tooth Disease Type 2D.

cell biology↗

bioRxiv2026

HPRC2: A human pangenome reference with near-complete coverage of common genetic variation

A pangenome reference overcomes the inherent limitation of any individual reference genome by integrating the variation present in a population. We present the Human Pangenome Reference Consortiums (HPRC) Release 2 (HPRC2), an openly available, second phase pangenome that is an approximately fivefold expansion in genome number over HPRC Release 1 (HPRC1) and measurable improvement in genome completeness, contiguity, and accuracy. Selecting samples with a principled algorithm prioritising common variant coverage, HPRC2 contributes 460 haplotypes that together capture over 99% of common variation observed in the All of Us Research Program v8 cohort. Combining high-coverage long and ultra-long reads with modern assemblers and polishers, we produce thousands of telomere-to-telomere (T2T) chromosomes, and relative to HPRC1 halve the number of structurally unreliable regions as well as individual base errors per haplotype. We complement the assemblies with whole genome multiple alignments and gene annotations, and derive formal pangenome coordinate systems for addressing off-reference variation, demonstrating that individual human genomes contain more than one hundred thousand variants not succinctly described with respect to existing reference genomes. We also present the first matched long-read backed pantranscriptome and panepigenome at this scale, provide continuous local-ancestry estimates spanning every genome, and outline a host of new tools and applications that leverage the pangenome resource for improved genomics analysis.

genomics↗

bioRxiv2026

Neuronal stop-codon readthrough is associated with ribosome pausing and alters protein localization in Drosophila

Stop-codon readthrough (RT) diversifies proteomes and is particularly prominent in neurons, suggesting its importance in nervous systems. However, the regulatory logic that specifies neuronal RT and the structural and cellular consequences of the resulting C-terminal protein extensions remain poorly understood. Here we leverage neuron-specific ribosome profiling datasets in Drosophila to construct an in vivo atlas of 163 neuronal RT transcripts. Sequence- based modeling distinguished RT from non-RT transcripts and highlighted an extended post-stop region enriched for stable predicted RNA structures. Beyond these cis-associated features, ribosome-footprint analysis revealed pronounced stop-codon pausing on RT transcripts, accompanied by upstream periodic peaks from the stop codon consistent with ribosome queuing. At the protein level, neuronal RT appended polypeptides enriched with polar residues and intrinsically disordered regions (IDRs). Finally, an in vivo dual-color reporter showed that RT of the RNA-binding protein Bru3 alters localization from the nucleus to cytoplasmic granules. Together, our results suggest that neuronal RT is associated with structured post-stop RNA regions that reshape termination dynamics and can produce IDR-rich C-terminal protein extensions with distinct subcellular localization.

molecular biology↗

bioRxiv2026

Pangenome-based human genome analysis improves trait association and genomic prediction

The Human Pangenome Reference Consortium has generated 462 open-access reference genomes and a variation graph that represents differences among them, providing a substrate for pangenome-based analysis methods that overcome the longstanding limitation of comparing all genomic data to a single linear reference. A key unresolved question is the extent to which these approaches can improve trait mapping. We investigate this using the genetics of gene expression variation as a model. We developed a graph-based method (EdgeDepth) for associating sequence variation with traits using short-read genome sequencing data, and show that it captures complex forms of genetic variation missed by other methods. We evaluated trait mapping performance using 430 samples with deep RNA-seq data, and found that pangenomic methods enable the detection of expression quantitative trait loci involving multiallelic indels and structural variants, leading to increased power at a subset of genes. These include 812 genes (7.9% of total) with [≥]20% improvement in statistical significance relative to the 1000 Genomes Project callset, and 185 (1.8%) with a 50% improvement, 10 of which are candidates to explain prior GWAS results. Notably, these analyses implicate GBAP1 pseudogene copy number as a causal factor in Crohns disease, likely via miRNA-mediated regulation of GBA1, which explains prior GWAS results based on flanking SNPs. The inclusion of pangenome-specific variation also improved the performance of gene expression prediction models, with median variance explained increasing from 10.1% to 12.5%, and 14.6% of genes showing significant improvement ({Delta}r2>0.05). Taken together, these results suggest that integration of pangenomic methods into human genetic studies will improve trait association and genomic prediction at a meaningful subset of genes.

genetics↗

bioRxiv2026

Anti-Polyamine Therapy Restrains Kidney Cyst Growth in an Orthologous Mouse Model of Autosomal Dominant Polycystic Kidney Disease

Autosomal Dominant Polycystic Kidney Disease (ADPKD) is the most common monogenic kidney disease worldwide. Tolvaptan is currently the only approved intervention to slow kidney cyst growth, but its indication is limited to rapidly progressing ADPKD and use compromises quality of life. Hence, new therapeutic options are of high clinical importance. Metabolic dysregulation is a hallmark of kidney cyst growth, including reprogramming of arginine metabolism and polyamine production. Sustained increases in polyamine synthesis drive pathological epithelial cell proliferation, tissue remodeling, and activation of M2-like macrophages, all known drivers of PKD. Difluoromethylornithine (DFMO) is an FDA approved irreversible inhibitor of ornithine decarboxylase (ODC1), the rate limiting enzyme in the arginine-polyamine pathway. Here, we treated C57Bl6/J p.R3277C (Pkd1RC/RC) mice for five months with DFMO-supplemented drinking water to test if DFMO treatment can slow kidney cyst growth in an orthologous model of ADPKD that mimics the disease pathophysiology seen in patients. At study end, Pkd1RC/RC mice treated with DFMO presented with significantly reduced percent kidney weight normalized to body weight as well as kidney cystic index and cyst number compared to control. Kidney RNA-seq analyses revealed correction of many genes and pathways perturbed in PKD post DFMO treatment, as well as overall heathier kidney cellular architecture as inferred by RNAseq-base cell type deconvolution. Our data highlight significant potential to repurpose DFMO for the treatment of patients with ADPKD and investigate the role of polyamines in modulating epithelial as well as myeloid cell fate in the setting of PKD.

pharmacology and toxicology↗

bioRxiv2026

Nature-inspired Nanoparticle Adiposomes Enable Targeted Delivery of Hydrophobic Drug for Anti-Cancer Treatment

The predominance of hydrophobic molecules in drug pipelines poses a fundamental delivery challenge in the aqueous physiological environment. Inspired by the endogenous lipid-transport machinery of lipid droplets and lipoproteins, adiposome was developed as a biocompatible platform for hydrophobic drug delivery. Here, docetaxel (DTX) was selected as a model hydrophobic chemotherapeutic agent to evaluate the therapeutic versatility of adiposome. Non-targeted DTX-loaded adiposomes (DTX-Ad) reduced systemic toxicity by avoiding the allergenic excipients Tween 80 and ethanol used in commercial DTX formulations. The cytotoxic mechanism and intracellular responses of DTX-Ad were comprehensively characterized. To boost antitumor efficacy, three tumor-targeting DTX-Ad formulations were subsequently engineered. Lung-targeting DTX-adiposomes (Lung@DTX-Ad), mediated by AAM-B1-28-CRGDK fusion peptide/NRP1 axis, enhanced antitumor efficacy in a lung metastasis model. BCMA-targeting DTX-adiposomes (BCMA@DTX-Ad), conferred by biotin-avidin-mediated antibody conjugation, achieved sustained suppression in a multiple myeloma model. Liver-targeting DTX-adiposomes (Liver@DTX-Ad), enabled by the ApoE/LDLR axis, outperformed both commercial DTX injection and sorafenib in hepatocellular carcinoma models. Taken together, these results demonstrate adiposome as a promising and broadly applicable drug delivery platform, with considerable potential for clinical translation across diverse malignancies.

cancer biology↗

bioRxiv2026

A Comprehensive Computational Model of the Human head for Designing and Optimizing Visual Brain-Machine Interfaces

Brain-machine interfaces (BMIs) for vision restoration require models that accurately simulate the anatomy and electrical properties of visual pathways. However, current models focus only on isolated structures, such as the retina or brain, and overlook surrounding tissues. Here, we present a comprehensive computational model of the human head, incorporating the entire visual pathway--including the eye, optic nerve, and brain--along with critical neighboring tissues such as the orbit, paranasal sinuses, enabling precise simulations. Validation using human and large animal data demonstrated a strong correlation between the simulated and measured electrical potentials. Component elimination analysis revealed that the optimized comprehensive model outperformed simplified versions. The models utility was demonstrated through multiple applications: (1) comparative analysis of electrical neuromodulation technologies for optic neuropathy, revealing the filed intensity limitations of noninvasive approaches and the safety concerns of invasive intraorbital approach; (2) identification of optimal stimulation site, revealing that transnasal stimulation at the optic chiasm outperformed traditional approaches; and (3) in silico design of electrode arrays for optic nerve prosthetics, demonstrating theoretical advantages in invasiveness and visual field coverage compared to existing retinal and cortical prosthetics. This validated and versatile computational resource supports the development of neuromodulation strategies and visual BMI technologies.

bioengineering↗

bioRxiv2026

Impacts of genome architecture on the repeatability of polygenic adaptation

A central but poorly resolved question regards how genome architecture shapes the selection response and repeatability of polygenic adaptation. We addressed this question using Evolve-and-Resequence experiments under rapid salinity decline in two sibling species (clades) of the invasive copepod Eurytemora affinis complex that differ strikingly in chromosome number (15 versus 4). The 4-chromosome genome arose from chromosomal fusions of the ancestral 15-chromosome genome, bringing together coadapted alleles at fusion sites. Across 10-20 generations of selection, both clades adapted to low salinity but followed divergent evolutionary trajectories. The selection lines of the 15-chromosome clade exhibited highly parallel genomic responses, whereas the 4-chromosome clade lines showed delayed and less repeatable responses. Forward genetic simulations revealed that strong synergistic epistasis among beneficial alleles best explained elevated parallelism in the 15-chromosome clade. Additional simulations varying chromosome number and epistasis revealed that strong positive epistasis, combined with high chromosome numbers, enhances parallelism by enabling recombination to assemble coadapted allelic combinations. In contrast, the high starting frequency of beneficial alleles in the 4-chromosome clade lines indicated selection on standing variation, likely acting on alternative linked haplotypes. These findings demonstrate that genome architecture and gene-gene interactions jointly determine the dynamics and predictability of polygenic adaptation.

evolutionary biology↗

bioRxiv2026

Beyond gene length: Exon-intron architecture and isoform potential in the evolution of eukaryotic complexity

Alternative splicing is a major source of human transcriptomic and phenotypic variation, yet its evolutionary contribution to genomic complexity remains unresolved. It has been shown that a mean gene length can be the most basic but remarkably efficient proxy for multicellular genome complexity, whereas mean protein length is not, as it plateaus abruptly early in eukaryotic evolution. Here, we show that, across 2,683 genomes, exon count continues to increase beyond this transition and then rapidly saturates at [~]10 exons per gene, supporting its role as an additional dimension of genomic complexity linked to exon-intron architecture. A simple stochastic exon-splitting model reproduces the observed biphasic exon-number growth pattern and identifies minimal exon length as a key determinant.

evolutionary biology↗

bioRxiv2026

Physiological perfusion of human vasculature reveals a YAP/TAZ-Apelin switch linking intraluminal flow to endothelial state transitions and vessel remodeling

Vascular flow delivers nutrients and imposes hemodynamic forces that govern vessel behavior in health and disease, yet fully human systems that recapitulate and tune physiological intraluminal flow in three-dimensional (3D) tissues are lacking. We developed VIVOS (Vascularized In Vitro Organ Systems), a platform that couples perfused human vascular beds to tunable pumps, generating continuous intraluminal flow through millimetre-scale vessels and 3D tissues at physiological shear stresses and pressures. VIVOS supports integration and perfusion of diverse human organoids and tissues, including lung organoids, cerebral organoids, vascular organoids, breast spheroids, and human retinal explants, as well as enables direct measurement and control of pressure, shear stress, and perfusion-dominant compound transport over extended culture periods. By tuning intraluminal flow and applying single-cell transcriptomics, we uncover a remodeling program in which laminar shear stress acts through a YAP/TAZ-TEAD "switch" to rewire an Apelin ligand-receptor axis and bias tip-stalk endothelial states, reshaping human vascular networks and linking hemodynamic cues to cell state transitions. We further model fast-flow arteriovenous malformations (AVMs) from Hereditary Hemorrhagic Telangiectasia and show that BMP9 constrains vessel caliber and perfusion while antagonizing a VEGF-driven angiogenic program, generating flow-quantified AVM-like lesions in a fully human 3D context. Together, these findings establish VIVOS as a generalizable platform that links physiological intraluminal flow to endothelial state transitions and vessel remodeling, enabling preclinical testing and mechanistic dissection of flow-regulated vascular pathologies in perfused 3D human tissues under defined hemodynamic conditions.

bioengineering↗

bioRxiv2026

FishMamba-1: A Linear-Complexity Foundation Model for Deciphering Polyploid Cyprinid Genomes

The Cypriniformes order, comprising essential aquaculture species like carps and minnows, presents unique genomic challenges due to complex whole-genome duplication (WGD) events and extensive repetitive elements. Conventional annotation tools and Transformer-based foundation models often struggle to capture long-range dependencies in these expanded genomes due to quadratic computational complexity. Here, we introduce FishMamba-1, the first genomic foundation model tailored for the aquatic clade, built upon the selective state-space model (SSM) architecture. By leveraging Mamba-2s linear scaling efficiency, FishMamba-1 processes context windows of 32,768 base pairs (32k)--significantly surpassing the 4-6k limit of standard DNA Transformers--enabling the modeling of distal regulatory patterns on a single GPU. We curated Cypri-24, a comprehensive dataset comprising 28.8 Gb of high-quality genome assemblies from 24 representative species, to pre-train FishMamba-1 on 15 billion tokens. Subsequent fine-tuning for genome segmentation (FishSegmenter) demonstrates the models capability to annotate gene structures at single-nucleotide resolution with remarkable precision. Evaluation on a held-out test set reveals that FishMamba-1 achieves a precision of 64.6% in exon identification, effectively distinguishing coding regions from the vast non-coding background without relying on RNA-seq evidence. Furthermore, interpretability analysis confirms that the model captures biological syntax such as splice acceptor motifs. FishMamba-1 provides a scalable, open-source framework for decoding the complex genomes of non-model organisms, providing a scalable computational resource to support downstream applications in molecular breeding and ecological monitoring. The complete source code, pre-trained model weights, and datasets are freely available at https://github.com/lu1000001/FishMamba. Additionally, the FishMamba Hub, a web-based inference platform, is accessible at https://huggingface.co/spaces/lu1000001/FishMamba-Hub to facilitate real-time genomic segmentation for the aquatic research community. Graphic Abstract O_FIG O_LINKSMALLFIG WIDTH=200 HEIGHT=112 SRC="FIGDIR/small/710409v1_ufig1.gif" ALT="Figure 1"> View larger version (71K): org.highwire.dtl.DTLVardef@f5ee02org.highwire.dtl.DTLVardef@1dcfd31org.highwire.dtl.DTLVardef@1729ef8org.highwire.dtl.DTLVardef@22e4f1_HPS_FORMAT_FIGEXP M_FIG C_FIG

bioinformatics↗