Biology subjects

Xu, Y.

Publications and source records attributed to Xu, Y..

At least 73 records · Page 4Linked to original sources

bioRxiv2026

Lexical and Semantic Coding in the Occipital Cortex of Sighted and Blind

How early visual cortex (EVC) supports language and semantic processing in sighted and congenitally blind individuals remains debated. Some predict that blindness induces radical functional reorganization of EVC, whereas others suggest more modest scaling-up changes of a neurofunctional architecture present in sighted individuals. We recorded whole-head MEG in 19 early blind (EB) and 21 sighted adults (SC) listening to spoken adjectives belonging to different semantic categories (i.e., abstract, concrete visual, concrete non-visual) and performing lexical and semantic decision tasks. In the lexical task, lexical information (i.e., word vs pseudoword) could be reliably decoded in perisylvian language areas and the EVC of both groups, with group differences (EB > SC) localized to occipital areas between 0.4 and 0.8 s after word onset. During the semantic task, semantic category information could be decoded in a network overlapping with the canonical semantic system and encompassing the EVC of both groups, with group differences (EB > SC) localized to occipital areas between 0.9 and 1 s after stimulus onset. We further characterized EVC properties and showed that i) EVC semantic information decoding is task sensitive, with no above-chance decoding in EB or SC in the lexical task ii) in the blind EVC it was possible to decode abstract from concrete concepts but not visual from non-visual concepts, a profile similar to that of posterior cingulate areas, but different to anterior cingulate cortex which showed above chance classification for all semantic categories. These results suggest that deprived EVC is integrated into a distributed lexical-semantic network carrying behaviourally relevant semantic information, although to a different extent.

neuroscience↗

bioRxiv2026

A hierarchical clock-mixture model for Bayesian phylogenetic dating

Conditioning the inference of a Bayesian phylogenetic time tree on a single molecular clock model treats clock choice as fixed, even when support among plausible clock families is uncertain. When that assumption is wrong, estimated timescales and their uncertainty can be distorted. Existing practice usually addresses this by fitting strict, uncorrelated lognormal relaxed (UCLN), and autocorrelated clocks separately and comparing their marginal likelihoods, but this requires multiple computationally expensive model selection analyses. Here we introduce a hierarchical clock-mixture framework, implemented as the open-source RelaxClockAveraging package for BEAST 2, that separates two questions that are often conflated in clock comparison: whether branch-specific rate variation is needed at all, and, if it is, whether that variation is better described as uncorrelated or autocorrelated. The method averages analytically between strict and relaxed clocks at the top level and then compares UCLN and autocorrelated models within the relaxed class on a shared branch-rate vector, returning posterior probabilities for all three clock families together with model-averaged summaries for parameters shared across them. In stratified simulations, the generating clock family was retained in the 95% posterior model set in all replicates, while model-averaged estimates of the overall substitution rate, root age, and tree length remained accurate. On a DENV-4 benchmark, the mixture reproduced the higher-effort nested-sampling ranking of clock families while avoiding the extreme run-to-run variability of independent marginal-likelihood estimates. On empirical benchmarks, the method recovered strong support for the autocorrelated clock on the classical 31-taxon chloroplast rbcL data set. On an RSV-A G-gene data set it concentrated virtually all posterior support on UCLN while preserving the established RSV-A timescale. On a 245-taxon structured-coalescent H3N2 dataset it assigned most posterior mass to the relaxed-clock class, with support within that class concentrated on the autocorrelated family, and the inferred timescale remained consistent with the published estimate. These results show that clock-model support and downstream timescale sensitivity are related but not identical. Single-clock dating analyses can still be adequate, but fixing a clock model should be justified by posterior support rather than treated as a default assumption.

evolutionary biology↗

bioRxiv2026

Neural Representations of Ensemble Mean and Variance Across Visual Features

Humans can rapidly extract summary statistics, such as the mean and variance of visual features, to efficiently represent complex visual environments despite limits in attention and working memory. For example, when viewing a field of flowers, we can perceive the average colour and overall variability of the display without individuating each flower. Although ensemble perception is central to visual cognition, two fundamental questions remain unresolved. First, it is unclear whether ensemble statistics for features represented at different levels of the visual hierarchy rely on a common neural system or on separate feature-specific systems. Second, it remains unknown whether different summary statistics, such as mean and variance, rely on shared or dissociable neural mechanisms. Here, we used fMRI and multivariate pattern analysis to examine the neural representation of ensemble mean and variance across three visual features spanning the processing hierarchy: orientation (low-level), shape (mid-level), and animacy (high-level) (N = 24; two fMRI sessions). By combining whole-brain searchlight and ROI-based approaches, we found a graded division of labour between ventral and dorsal visual pathways. Although mean and variance ensemble statistics were distributed across the visual cortex, mean decoding was stronger in ventral regions, whereas variance decoding was stronger in dorsal regions. Ensemble mean representations followed a posterior-anterior gradient within the ventral visual pathway, consistent with increasing abstraction from orientation to shape and animacy, and showed little anatomical overlap suggesting largely feature-specific. By contrast, ensemble variance was weighted toward dorsal parietal and frontoparietal regions, especially superior parietal cortex and intraparietal sulcus, decoding clusters largely overlap across features and generalized robustly across orientation, shape, and animacy. Together, these findings provide a more nuanced account of ensemble perception, showing that feature-specific and feature-independent neural codes can coexist across visual cortex and help reconcile previously conflicting evidence. Significance StatementThis study shows that information about ensemble mean and variance is distributed throughout the visual cortex but weighted differently across visual pathways. Mean-related information was stronger in the ventral visual pathway and showed a largely feature-independent anatomical organization, whereas variance-related information was stronger in dorsal parietal regions and generalized across orientation, shape and animacy. Together, these findings provide clarifying neural evidence for longstanding debates about whether ensemble statistics rely on shared or dissociable mechanisms and whether ensemble representations are feature-specific or generalize across visual features. More broadly, the study offers a nuanced account of ensemble perception in which feature-specific and feature-general neural codes coexist across visual cortex, potentially reconciling previously conflicting evidence.

neuroscience↗

bioRxiv2026

Designing Fidelity of CRISPR-Cas Endonucleases by Kinetic Insights

Finding high-fidelity CRISPR-Cas variants is critical for both the precision of in vitro DNA detection and the safety of in vivo gene-editing therapeutics. However, the large size of the Cas enzyme and the distinct selection criteria between its natural evolution and clinical practice lead to extensive experimental trials and limited success rate for de novo design, directed evolution, protein language model (PLM)-based filtering. Here, we present a PLM-assisted physics-driven approach that utilizes atomistic molecular dynamics simulations and automated path searching to efficiently obtain the complete kinetic insights, including the transition state structures, for the conformational changes of Cas before DNA cleavage. We show that these kinetic insights can pinpoint a few fidelity-diminishing protein residues during the early stage of target recognition, and have led to SpyCas9 and FnCas12a variants with ultra-high fidelity surpassing previously reported counterparts at minimal cost of a few wet-lab trials.

biophysics↗

bioRxiv2026

Frontal Eye Field Leads a Distributed Oculomotor Circuit for Abstract Categorical Decisions

Flexible decisions require the brain to transform sensory evidence into abstract, task-relevant variables and then into actions. Understanding this process requires identifying how distributed neural populations represent sensory, cognitive, and motor variables, and how interareal interactions mediate transformations between them. We simultaneously recorded population activity in frontal eye field (FEF), lateral intraparietal area (LIP) and superior colliculus (SC) while monkeys performed a flexible yet urgent visual motion-categorization task. Within this network, FEF first encoded abstract categories, followed by SC and then LIP. LIP showed the earliest encoding of visual stimulus features, but a later encoding of upcoming saccades. Single-trial analyses revealed directed information flow from FEF to LIP populations for category-and choice-related signals. Reversible FEF inactivation impaired categorization and saccadic choice, causally implicating FEF in category-guided action. These findings reveal a differentiated FEF-LIP-SC circuit for transforming sensory evidence into abstract categorical decisions and the actions used to report them.

neuroscience↗

bioRxiv2026

CatESO: Differentiable Enzyme Sequence Optimization Guided by Substrate-Aware kcat Prediction

Enzymes drive biological chemistry and offer greener routes to chemicals, materials and medicines, yet their broader use as biocatalysts is often limited by insufficient catalytic turnover. Improving turnover is hard: measured rate constants are scarce and protein sequence space is vast. Deep-learning models now predict the turnover number, kcat, with growing accuracy, but they are typically applied after sequence generation to score or filter candidates, which separates the kinetic objective from the design itself. To bridge the gap between sequence generation and kinetic evaluation, we introduce CatESO, a differentiable sequence optimizer that enables direct, gradient-guided design of substrate-specific catalytic turnover. By backpropagating through a cross-modal kcat predictor under continuous sequence relaxation, CatESO co-optimizes predicted catalytic activity, evolutionary plausibility and structural integrity in one end-to-end framework, using ESM-2 and ESMFold to keep designs evolutionarily plausible and foldable. Across seven stringent out-of-distribution enzymes spanning EC classes 1-7, CatESO raised model-predicted kcat for the vast majority of designs, with a median predicted fold change of 1.52 while every variant retained a pLDDT above 70. Against RFdiffusion3-LigandMPNN pipeline and ZymCtrl, CatESO struck a better balance between predicted activity and structural confidence. By making substrate-conditioned kinetic objectives differentiable, CatESO carries differentiable protein design beyond structure- and binding-centred goals to enzyme catalytic function, giving a general route to function-oriented enzyme engineering.

biochemistry↗

bioRxiv2026

PIP2 stabilizes Nav1.5 gating and links receptor signaling to cardiac late sodium current

The cardiac sodium channel NaV1.5 initiates each heartbeat by generating the rapid depolarizing upstroke of the action potential. Dysregulation of NaV1.5 gating can produce cardiac arrhythmias by slowing inactivation, increasing late sodium current (INa,L), and impairing electrical stability. Here, we show that phosphatidylinositol-4,5-bisphosphate (PIP2) is a critical membrane cofactor that stabilizes NaV1.5 gating. Acute PIP2 depletion in human iPSC-derived cardiomyocytes, produced by activation of endogenous AT1 receptors, activation of an engineered M3q-DREADD, or optogenetic recruitment of CRY2-pseudojanin, shifted voltage dependence, slowed fast inactivation, and increased INa,L. These effects were prevented by augmenting intracellular PIP2, required PLC activity when driven by Gq-coupled receptors, and were independent of downstream Ca{superscript 2} or PKC signaling. Unlike the skeletal-muscle isoform NaV1.4, NaV1.5 displayed PIP2-dependent shifts in both activation and steady-state inactivation, indicating isoform-specific lipid coupling. Induced-fit docking and molecular dynamics simulations identified a PIP2-interaction interface between the domain IV voltage sensor and pore that contains disease-linked residues. The disease-reported variant R1644C weakened and redistributed the predicted PIP2-contact network, produced elevated basal INa,L, showed enhanced sensitivity to PIP2 depletion, and caused an approximately 30-fold reduction in apparent functional PIP2 sensitivity in excised patches. These findings define a lipid-dependent mechanism that stabilizes NaV1.5 gating and reveal how physiological Gq signaling and inherited channel variants can converge on the channel-PIP2 axis to promote proarrhythmic late sodium current.

physiology↗

bioRxiv2026

Dense CRISPR Mapping of the WAVE Complex Identifies Pharmacological Modulators of AD-Linked Myeloid Migration

Although genetic risk for Alzheimers disease (AD) strongly converges on microglial pathways, the druggable functional protein regions that control disease-relevant microglial behaviors remain largely unknown. Here, we applied dense CRISPR-Cas9 mutagenesis and CRISPRtile-based functional mapping to the WAVE regulatory complex (WRC), a central regulator of actin remodeling and cell migration. In a pooled CCL2-directed migration assay in human THP-1 myeloid cells, perturbation of NCKAP1L, CYFIP1, and BRK1 impaired migration and revealed divergent effects among WRC paralogs. Residue-level analysis mapped discrete migration-associated functional regions within CYFIP1 and NCKAP1L, including a CYFIP1 regulatory hotspot and a prioritized NCKAP1L region nominated for pharmacological targeting. Human single-nucleus datasets identified NCKAP1L as a microglia-enriched WRC component expressed across diverse microglial states. Machine-learning-guided compound prediction nominated Montelukast sodium and Piperacetazine, which we experimentally validated as negative and positive modulators of chemokine-directed migration, respectively. These findings establish WRC-dependent migration as a pharmacologically tunable myeloid process relevant to AD.

neuroscience↗

bioRxiv2026

StructureSAFE: A structure-aware chemical language model for unified hit identification and lead optimization

Structure-based generative models (SBGMs) hold great promises for accelerating drug discovery by enabling target-aware molecular design. However, existing approaches face fundamental challenges: three-dimensional graph-based models can explicitly incorporate protein structural information but often generate chemically implausible molecules due to limited training data, while chemical language models (CLMs) produce chemically plausible molecules but struggle to effectively leverage three-dimensional structural information for structure-conditioned generation and hard to incorporate lead optimization functionality due to the nature of SMILES string. Here, we present StructureSAFE, a structure-aware chemical language model that resolves this trade-off by integrating protein structural and evolutionary encoders with the SAFE molecular representation via pretraining and finetuning training scheme, enabling both de novo hit identification and a comprehensive suite of lead optimization subtasks within a unified framework. Comprehensive benchmarking on the MolGenBench dataset demonstrates that StructureSAFE achieves state-of-the-art (SOTA) performance across multiple metrics, with particularly pronounced improvements in chemical plausibility relative to graph-based models lacking pretraining. Evaluation on a rigorously constructed held-out test set further confirms its ability to generate drug-like, synthetically accessible molecules with competitive predicted binding affinities for previously unseen targets on both hit identification and lead optimization setting. In silico case studies across four therapeutically relevant targets validate its capacity to generate chemically plausible molecules that recapitulate key binding interactions of known high-affinity ligands while proposing novel interactions for potential better affinity and exploring previously unknown regions of chemical space. Taking together, StructureSAFE represents a versatile and practical tool to provide high-quality candidate molecules for augmenting medicinal chemistry workflows in both hit identification and lead optimization campaigns.

bioinformatics↗

bioRxiv2026

A Hypothalamic Inhibitory Circuit Encoding the Scalability of Stress Responses

An appropriate stress response is essential for properly responding to, coping with, and subsequently recovering from disturbing environmental stimuli. However, how the brain dynamically encodes the scalability of stress responses remains poorly understood. Here, we found that, GABAergic neurons in the arcuate nucleus (Arc, denoted as ArcGABA neurons) send direct inputs to corticotropin-releasing hormone (CRH) neurons in the paraventricular nucleus of the hypothalamus (PVH, denoted as PVHCRH neurons), the primary regulators of the hypothalamic-pituitary-adrenal (HPA) axis. Although PVHCRH neurons exhibited time-locked activation in response to various environmental stressors, both GABA release onto PVHCRH neurons and the activity of PVHCRH-projecting ArcGABA neurons were selectively reduced during exposure to prolonged, high-intensity stressors, but not following exposure to transient, low-intensity stressors. Notably, GABA release onto PVHCRH neurons was positively correlated with PVHCRH-projecting ArcGABA neuron activity, yet anticorrelated with PVHCRH neuronal activity in response to the same prolonged, high-intensity stressors. Selective silencing of PVHCRH-projecting ArcGABA neurons was sufficient to elevate HPA axis activity and stress levels, phenocopying the effect of direct of PVHCRH neuron activation. Conversely, selective activation of PVHCRH-projecting ArcGABA neurons reduced both HPA axis activity and stress levels, this effect was completely abolished by concurrent excitation of PVHCRH neurons. Molecular identity screening further revealed that these PVHCRH-projecting ArcGABA neurons are not subsets expressing agouti-related peptide (AgRP) and tyrosine hydroxylase (TH) markers. Collectively, these findings indicate that the non-AgRP/TH ArcGABA PVHCRH neurocircuit serves as a critical neural substrate that directly encodes the scalability of stress responses to environmental stressors by modulating inhibitory GABA release in a stimulus intensity-dependent manner.

neuroscience↗

bioRxiv2026

Quantitative Motion-Corrected PALM Links Endosome Structure and Dynamics in Live Cells

Quantitative structural analysis by Photoactivated Localization Microscopy (PALM) on the nanoscale is often restricted to fixed cells because motion during prolonged data acquisition distorts image reconstruction. Here, we develop motion-corrected PALM (mcPALM), a live-cell super-resolution approach combining a conventional fluorescence channel with PALM to correct motion-induced spreading of localizations. We further introduce a photoactivation-based correction to estimate molecule numbers from incomplete trajectories. Using PI3P-marked endosomes in yeast as a dynamic model system, we show that mcPALM recovers a live-cell maturation trajectory linking motion-corrected endosome size and calibrated PI3P content, consistent with fixed-cell benchmarks. Unlike fixed-cell PALM, mcPALM preserves endosome dynamics, revealing stage-dependent directed transport and maturation-associated motility shift. Thus, mcPALM extends PALM from static structural measurements in fixed samples to integrated quantification of nanoscale structure, molecular composition and dynamics in living cells. This framework is broadly applicable to other mobile organelles and biomolecular assemblies, enabling live-cell studies on how molecular organization and dynamics are coupled to biological function.

biophysics↗

bioRxiv2026

Cardiovascular Benefits of Menopause Hormone Treatment is Age-Dependent

BackgroundHormone therapy (HT) has not consistently reduced atherosclerotic cardiovascular disease (ASCVD) events in post-menopausal women, yet the underlying mechanisms remain poorly understood. MethodsFemale Ldlr-/- mice with established atherosclerosis were subjected to surgical menopause and treated with 17{beta}-estradiol (E2) following lipid normalization. Studies were performed in aging and young mice. To determine whether inflammation mediates the age-dependent response to HT, a cohort of aging mice underwent transplantation with Ifn{gamma}-/- bone marrow (BM) before hormone treatments. Metabolic parameters, HDL function, systemic inflammation, atherosclerotic burden, liver metabolic and oxidative stress signaling, and hepatic estrogen receptor signaling were evaluated. ResultsIn aging mice, menopause E2 treatment failed to reduce established atherosclerosis as shown in sham operated mice during lipid normalization. Instead, E2 treatment increased circulating IFN{gamma} and IL-6, impaired HDL antioxidant and cholesterol efflux functions, and promoted inflammatory and vulnerable plaque phenotypes. Suppression of inflammation through Ifn{gamma}-/-BM transplantation restored HDL function and significantly reduced atherosclerosis in E2-treated aging mice. In contrast to aging mice, young mice exhibited reduced systemic and plaque inflammation, improved HDL functions and atherosclerosis following E2 treatment. Liver RNA sequencing and qPCR validation identified activation of inflammatory, oxidative stress, and lipid metabolic pathways in aging E2-treated mice, which were largely attenuated following Ifn{gamma}-/-bone marrow transplantation as well as in young mice. Compared to young mice, aging mice presented hepatic estrogen receptor remodeling characterized by reduced estrogen receptor (ER) expression and increased G-protein coupled estrogen receptor (GPER) expression. Constitutive GPER activation was accompanied by induction of NOX1-dependent oxidative stress, which was further exacerbated by E2 treatment, leading to persistent inflammation. ConclusionsThe cardiovascular effects of estrogen therapy are fundamentally age dependent. Aging shifts estrogen signaling toward hepatic oxidative stress and inflammation through increased GPER. While E2 treatment preserves both metabolic and cardiovascular protection in young mice, aging exacerbates GPER-NOX1-mediated oxidative stress, resulting in impaired HDL function and persistent residual ASCVD risk. These findings identify inflammation-driven, non-lipid mechanisms as potential therapeutic targets to improve cardiovascular outcomes during hormone therapy in postmenopausal women.

physiology↗

bioRxiv2026

NLR from soybean Rsv1 locus confers broad-spectrum resistance to soybean mosaic virus G1-G7 strains by recognizing viral P3 protein

Nucleotide-binding leucine-rich repeat (NLR) immune receptor genes are of significant value in disease resistance breeding and the control of viral diseases. Soybean mosaic virus (SMV) poses a serious threat to soybean production and the Rsv1 locus in soybean cultivar Suweon 97 confers broad-spectrum resistance against SMV strains G1 to G7; however, this locus harbors no fewer than 18 NLR genes, and thus the broad-spectrum antiviral mechanisms underlying the Rsv1 locus remain poorly understood to date. Here, we established a rapid and highly efficient screening system for cloning NLR genes from soybean Rsv1 locus and identified a broad-spectrum antiviral NLR gene 13g184900 from this highly complicated locus. The NLR encoded by 13g184900 can recognize viral P3 protein from all SMV strains (G1-G7) and another potyvirus Bean common mosaic virus (BCMV). The coiled-coil (CC) domain of this NLR directly interacts with viral P3 protein. Additionally, we showed that this NLR originated from wild soybean accession in East China and has been introduced into several soybean cultivars during domestication. Collectively, we developed a high-throughput screening system for identifying NLR genes in soybean and our study provides new mechanistic perspective on how the Rsv1 locus mediates the broad-spectrum resistance to all SMV G1-G7 strains.

plant biology↗

bioRxiv2026

Circadian temperature compensation is intrinsically linked with metabolism and redox signaling

Circadian rhythms are a universal property of organisms, and a defining property of these oscillators is the ability to maintain a precise period at different temperatures. Indeed, the discovery in the 1950s of this property, known as "temperature compensation," was a watershed event in our understanding of biological timekeeping. Since that time, however, almost no general principles have emerged that uncover the mechanistic basis of how circadian timekeepers defy the rules" of temperature dependence of biochemical reactions. A genetic investigation of circadian temperature compensation revealed that the sensing of metabolism and ensuing modulation of the core mechanism preserves a constant circadian period. Intracellular redox sensing of metabolic rate is intrinsic to this compensation, and this relationship is conserved in bacterial and mammalian cells. These insights explain previously inexplicable observations of the interaction between redox reactions and circadian timekeeping. These results introduce a previously unknown linkage between temperature compensation and metabolism.

cell biology↗

bioRxiv2026

Discover Novel RNA Targeting Small Molecules by Fluorescent Aptamer Screening

Discovering small molecules targeting proteins represents a major effort in drug development. RNA, however, as a class of macromolecule that carrying out important regulatory roles in the cell as drug target, only received attention recently. Although several methods have been proposed, an easy to operate, fast and robust method is still lacking. We designed a generic florescence screening method by fusing the target RNA with a florescent aptamer (fusion RNA) and then carried out screening using high-throughput format (Fluorescent Aptamer Screening, FAS). In this work, we chose SL5 on SARS-Cov-2 5UTR as the test target. SL5 is a conserved motif across several corona virus family members whose core is not prone to mutation. We screened 9528 compounds, successfully identified four molecules (Sertraline (hydrochloride), Samuraciclib (hydrochloride), Minocycline (hydrochloride), JG-98 bind direct to the full-length SL5 at micromolar or higher affinity. The design of FAS could be easily adapted to structured RNA motifs without prior knowledge of its 3D structural information. In addition, this work showed the possibility of developing generic drugs for RNA virus by targeting the conserved viral RNA genome and paved a new way for the discovery of small molecule drugs in combating human diseases.

Preprint↗

bioRxiv2026

A post-mitotic bistable switch actively safeguards mature neurons against reciprocal transdifferentiation in vivo

How mature cells actively defend their identity against alternative fates is poorly understood. Here we show that two types of fully differentiated Drosophila somatosensory neurons, derived from distinct lineages, are continuously prevented from transdifferentiating into one another by a post-mitotic bistable switch. Two homeodomain transcription factors, Hmx and Cut, are co-expressed in both neuronal types but adopt opposite stoichiometric dominances through mutual repression, each defining one identity. Disrupting this balance in mature neurons triggers deterministic, lineage-crossing reprogramming of transcriptome, dendritic and axonal architecture, and sensory behavior without reversion to a progenitor state. This regulatory logic is evolutionarily conserved: Drosophila sensory identities share transcriptomic signatures with distinct human A{delta} mechanoreceptors, and human CUX and HMX proteins functionally replace their fly counterparts. Mature neuronal identity is thus not a fixed endpoint but a poised equilibrium actively maintained by ongoing competition between opposing selectors. TEASERMature somatosensory identity is a poised equilibrium actively maintained by continuous competition between opposing terminal selectors.

Developmental Biology↗

bioRxiv2026

scIsoAgent enables autonomous isoform-resolved characterization and sequence-informed interpretation of long-read single-cell transcriptomes

Alternative isoform usage can alter gene function independently of total gene expression, creating a need to resolve transcript isoforms at single-cell resolution. Long-read single-cell RNA sequencing meets this need by linking cellular identity to transcript isoforms and sequence-level features. Realizing its full biological value requires reproducible workflows that connect specialized long-read analysis with biological interpretation. Existing large language model (LLM)-based biomedical agents support general omics analysis, but are not designed for isoform-resolved long-read single-cell workflows. Here, we present scIsoAgent, an autonomous LLM-powered scientific agent for long-read single-cell RNA-seq analysis. scIsoAgent turns heterogeneous long-read single-cell inputs into traceable isoform-resolved workflows, using stage-aware planning and persistent computational context to support both execution and interpretation. Across complementary evaluations, this design improved the continuity from analysis planning to executable, interactive workflows compared with general-purpose LLM baselines. In real-data reanalysis, scIsoAgent recovered major findings from published long-read single-cell resources and extended a representative differential transcript usage event into a sequence-informed functional hypothesis. By linking full-length isoform sequences with model-inferred transcript properties, scIsoAgent connects observed isoform usage with potential sequence-level functional consequences. These results demonstrate that autonomous scientific agents can transform fragmented long-read single-cell analysis into coherent, reproducible workflows for isoform-resolved discovery and biological interpretation.

bioinformatics↗

bioRxiv2026

Structure of the mammalian ciliary transition zone microtubule doublet

The ciliary transition zone gates bidirectional protein trafficking to maintain the specialized ciliary proteome using microtubule doublets as a scaffold. While ciliary axonemal doublets are well-characterized, the molecular architecture of the transition zone doublet remains elusive. Here, we report the structure of the mammalian transition zone doublet from bovine tracheal cilia using cryo-electron tomography at 4.7-5.0 [A] resolution. The transition zone doublet is a structurally independent segment defined by an 8 nm-periodic arrangement of unique microtubule-inner proteins (MIPs) and microtubule-associated proteins (MAPs). We identify the calcium-binding protein CAPSL as a lumenal MIP that forms a pseudo-helical spiral and stabilizes microtubules in vitro. Furthermore, a dense MAP network on the A-microtubule surface clashes with intraflagellar transport (IFT) motor binding sites, suggesting anterograde IFT is directed to the B-microtubule for potential regulation by MAP9. Our work provides a structural framework for understanding gated ciliary transport and transition zone-linked human ciliopathies.

Cell Biology↗