Biology subjects

Chen, J.

Publications and source records attributed to Chen, J..

At least 91 records · Page 5Linked to original sources

bioRxiv2026

A Conversational Multi-Agent AI System for Integrated Multi-Omics Analysis and Biomedical Discovery

Single-cell and spatial omics offer unprecedented opportunities to decipher the mechanisms of disease, however, this process requires teams of experts, iterative trial-and-error and reasoning across modalities. Here we present LungChat (https://chat.lungmap.net), a conversational system for integrated multi-omics analysis and biomedical discovery, deployed as a hierarchical multi-agent architecture in which a supervisor decomposes natural-language questions into parallel, tool-grounded tasks spanning single-cell and spatial analyses, literature and clinical-trial synthesis, and drug repurposing. To predict new therapeutics, LungChat implements Direction-Aware Repurposing and Targeting (DART) to distinguish perturbations that reverse disease transcriptional programs from those that reinforce them, at the cell-type level, for safety prediction. Controlled architecture ablations showed that hierarchical orchestration improved grounded abstention and token efficiency and preserved strong performance on complex multi-step tasks. In pulmonary disease case studies, LungChat independently prioritized saracatinib for IPF through drug-connectivity screening, followed by DART-based cell-type analysis; the same compound has been evaluated in the STOP-IPF clinical trial (NCT04598919). The system also recovered fluticasone propionate, an established COPD therapy, through a single orchestrated analysis. This tissue-agnostic system provides a blueprint for verifiable agentic AI systems that support reproducible scientific discovery.

bioinformatics↗

bioRxiv2026

dbverse scales spatial omics analysis with embedded analytical databases

Spatial omics datasets are increasing in size and complexity, exceeding the memory of standard computers and thereby limiting data analysis. Here we present dbverse, a framework for larger-than-memory matrix, spatial and genomic data analysis in embedded analytical databases. Benchmarks show dbverse provides orders of magnitude runtime improvements relative to established in-memory and file-backed methods for core operations in single-cell and spatial omics analysis. We integrated dbverse with Giotto Suite, scaling end-to-end preprocessing of millions of cells and enabling spatial alternative polyadenylation analysis as demonstrated on a Visium HD 3' ovarian clear cell carcinoma sample. The dbverse framework provides an interoperable database foundation for larger-than-memory spatial omics analysis on ordinary computers.

bioinformatics↗

bioRxiv2026

SpatialMOC: Accurate reconstruction of spatial multi-omics landscapes through cross-modality prediction

Spatial multi-omics technologies provide unprecedented opportunities to characterize tissue organization by measuring complementary molecular layers within their native spatial context. However, simultaneous profiling of multiple molecular modalities remains technically challenging, limiting the widespread application of spatial multi-omics and leaving most studies reliant on single-modality measurements. Here, we present Spatial Multi-Omics Cross-prediction (Spatial-MOC), a computational framework that reconstructs missing spatial molecular modalities by integrating spatial context with cross-modality representation learning. Across multiple tissues, molecular modalities, developmental stages, sequencing platforms, and degraded datasets, SpatialMOC consistently outper-formed existing computational approaches in bidirectional molecular prediction. Beyond accurate prediction, SpatialMOC faithfully reconstructed tissue architecture, preserved dynamic molecular and regulatory heterogeneity, and recovered biologically meaningful spatial landscapes from technically compromised measurements. Together, these results establish SpatialMOC as a general framework for spatial multi-omics reconstruction, extending the analytical value of existing spatial omics datasets and facilitating comprehensive investigations of tissue organization, development, and disease.

bioinformatics↗

bioRxiv2026

A cross-modal generative model for incomplete and degradedprostate MRI with multicentre clinical validation

Missing or degraded sequences can limit prostate multiparametric MRI. We developed MSCNet, a sequence-conditioned cross-modal generative framework for reconstructing unavailable contrasts and restoring degraded acquisitions. Across ten completion tasks, task-specific MSCNet achieved mean structural similarity of 0.818 versus 0.798 for the strongest task-matched comparators; matched-capacity analyses showed larger differences in lesion fidelity and boundary preservation. In a blinded 1,000-case reader study, overall image quality met the prespecified non-inferiority criterion for DWI, ADC and T2W completion, but not T1W. In a separate 200-case diagnostic assessment, AUCs for clinically significant cancer were 0.860 with acquired images, 0.841 with MSCNet and 0.797 with baseline-generated images. A locked 186-case three-hospital cohort supported multicentre transportability. These retrospective results support quality-controlled cross-modal reconstruction as an adjunct to acquired prostate MRI.

bioengineering↗

bioRxiv2026

Overexpression of miR-424(322)/-503 induces severe dilated cardiomyopathy by regulating the fatty acid oxidation gene expression program

Fatty acid oxidation (FAO) is a major energy source in the adult heart, and disruption of cardiac metabolism is closely associated with heart failure. Here, we investigated the effects of cardiac-specific overexpression of the H19X-encoded miR-424(322)/-503 cluster using an inducible transgenic mouse model. Sustained miR-424(322)/-503 overexpression caused rapid metabolic and functional deterioration, with early impairment of fatty acid oxidation. Short-term induction reduced FAO activity and downregulated genes involved in lipid metabolism, while glycolytic enzyme activity remained largely unchanged. Continued miR-424(322)/-503 expression subsequently led to severe dilated cardiomyopathy characterized by ventricular dilation, wall thinning, fibrosis, reduced contractility, and high mortality. Importantly, disease progression was dependent on the level and duration of miR-424(322)/-503 expression, as intermittent or lower-dose induction delayed cardiac dysfunction and prolonged survival. Withdrawal of miR-424(322)/-503 expression after the onset of dysfunction promoted substantial functional recovery. Together, these findings identify miR-424(322)/-503 as a potent regulator of cardiac metabolic reprogramming that disrupts fatty acid metabolism and drives progressive heart failure.

molecular biology↗

bioRxiv2026

Antibiotic condensates activate magnesium signaling to drive drug tolerance

The spatial distribution of small molecules within cells shapes their biological activity, yet these distributions are generally assumed to be governed passively by reaction-driven electrochemical gradients. Here we show that aminoglycoside antibiotics actively control their own subcellular organization by undergoing phase separation with RNAs. Combining in vitro reconstitution, bacterial assays, and molecular dynamics simulations, we discovered that aminoglycosides coacervate with RNA through multivalent electrostatic interactions, displacing and releasing RNA-bound Mg2+. This condensate-dependent Mg2+ release remodels the cytosolic labile Mg2+ pool and activates magnesium signaling. This effect dampens the magnesium-starvation regulation, sustains ribosome activity, and shifts the cellular electrochemical state, promoting bacterial fitness. Because condensation occurs only above a defined concentration threshold, it generates a non-monotonic dose-response in which higher antibiotic concentrations paradoxically enhance bacterial survival. This antibiotic condensate-dependent Mg2+ signaling confers tolerance to multiple ribosome-targeting antibiotics simultaneously even in cells lacking resistance gene, while condensate dissolution restores antibiotic efficacy. Our findings establish antibiotic-driven phase separation as a previously unrecognized mechanism to encode cellular signaling and identify antibiotic condensates as a distinct functional unit underlying drug tolerance.

microbiology↗

bioRxiv2026

Trustworthy in silico labeling via semantic visual interpretability of image-to-image translation

Cross-modality image translation promises to provide multiple layers of biological information from a single image input, yet its practical application is stalled by a lack of interpretability and the inability to account for model imperfections. In silico labeling, the inference of organelle localization from label-free images, is a primary example where this black-box nature limits adoption. We present Mask Interpreter, a generalized method for semantic visual interpretability of image-to-image translation models. By uncovering organelle-specific "explanation signatures", Mask Interpreter validates that models rely on authentic biological structures rather than spurious artifacts. Beyond biological validation, it outperforms traditional explainable AI (xAI) approaches, identifies batch effects and localizes prediction errors when ground-truth fluorescence is unavailable. Semantic confidence modeling further provides fine-grained reliability assessment at single-cell resolution, enabling the automated exclusion of artifacts from downstream analyses. By bridging the gap between computational inference and meaningful biological features, Mask Interpreter transforms in silico labeling into a reliable tool for scientific discovery across diverse biomedical imaging modalities.

cell biology↗

bioRxiv2026

Phosphorylation and pocket-binding ligands rewire the coupled networks in multidomain protein condensate

Biomolecular condensates formed by multidomain proteins are increasingly linked to disease and are emerging targets for chemical modulation. Yet most molecular models emphasize intrinsically disordered regions (IDRs), whereas many available ligands bind folded-domain pockets. How IDR modifications and folded-domain ligands are integrated to control condensate organization remains unclear. Here, we address this problem using histone deacetylase 6 (HDAC6), a disease-associated deacetylase whose S22 phosphorylation and catalytic-pocket ligands differentially regulate condensation. Using atomistic-informed Martini 3 coarse-grained simulations, we show that S22 phosphorylation enhances phase separation of the N-terminal IDR1 without compacting isolated chains. Instead, phosphorylation creates a phospho-S22/D26-centered anionic contact node that strengthens dense-phase interactions with Arg/Lys-rich patches. The HDAC6 ligands Nexturastat A and HPOB remodel this IDR network through distinct charged-contact mechanisms: Nexturastat A acts as a focused competitor for cationic patches, whereas HPOB permits partial compensation through alternative charged contacts. In full-length phospho-HDAC6, the assembled network is not IDR1-dominated but is organized by FD3, the C-terminal ZnF-UBP domain. Pocket-bound ligands redirect this folded-domain network from FD3-FD3 association toward pocket-FD3 contacts, with stronger and more persistent remodeling by Nexturastat A. These results reveal a two-layer regulatory mechanism in which phosphorylation rewires IDR electrostatics while pocket-binding ligands redirect folded-domain interactions, suggesting a strategy for modulating multidomain condensates through ligand-induced contact-network remodeling.

biophysics↗

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

Closed-loop optical optimization enables patterned retinal stimulation in vivo at cellular scales

Visual neuroscience requires precise spatiotemporal projection of optical stimulation onto the retina, especially in experimental mouse models. However, in vivo patterned stimulation in mice is profoundly hindered by the extreme optical power and severe anatomical aberrations of the eye. Consequently, visual stimulation relies mainly on unverifiable, open-loop approximations that often lack spatial precision. Here, we introduce a closed-loop, spatially modulated stimulation platform that overcomes these barriers. By integrating a digital micromirror device (DMD) with electronically tunable lenses (ETLs) and a real-time, fundus camera-guided focus optimization module, we directly verify the location of patterned stimuli on the retina while dynamically correcting for chromatic and geometric defocus. This platform delivers quantitatively verified static and dynamic patterned stimuli to the living retina with lateral resolutions as fine as 6.7 {micro}m. Guided by ray-tracing optical analysis, our work establishes a technological foundation that enables highly reproducible, cellular-scale interrogations of the visual pathway.

bioengineering↗

bioRxiv2026

Pseudouridines in the U2 Branch Site Recognition Region Differentially Contribute to Branch Site Recognition during Pre-mRNA Splicing

Pseudouridines ({Psi}s) are highly enriched in conserved, functionally critical regions of spliceosomal snRNAs, particularly within the U2 branch site recognition region (BSRR), which contains six {Psi}s in humans and three in S. cerevisiae. To investigate how U2 BSRR pseudouridylation influences branch site sequence (BSS) recognition, we developed a large-scale, high-throughput screening system in S. cerevisiae that allows a library of pre-mRNAs with randomized BSSs to be spliced in distinct U2 BSRR pseudouridylation backgrounds. Screening and next-generation sequencing (NGS) revealed that different U2 pseudouridylation backgrounds exhibit distinct recognition patterns and efficiencies for specific BSSs. Notably, {Psi}42 and {Psi}44 generally enhanced splicing, whereas {Psi}38 alone, and in some contexts together with {Psi}35, consistently impaired BSS recognition. The differential effects were validated using endogenous S. cerevisiae genes. In addition, splicing assays with engineered pre-mRNA constructs guided by the screening results demonstrated that BSRR {Psi}s directly influence BSS selection, supporting a model in which U2 pseudouridylation modulates BSS recognition and could contribute to alternative splicing in more complex eukaryotes. Finally, synthetic-lethality analyses with a Prp5 mutant, together with Prp5-U2-pre-mRNA binding assays, indicate that U2 BSRR {Psi}s are critical for Prp5 recruitment and, consequently, for proper U2-BSS interactions during spliceosome assembly. Collectively, these findings establish U2 pseudouridylation as a key determinant of branch site recognition and spliceosome function.

molecular 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

Bone Lymphatic Endothelial Cells Form a Distinct Skeletal Population and Drive Skeletal Repair

Bone regeneration relies on specialized vascular niches, yet the contribution of lymphatic vessels across distinct skeletal sites remains poorly understood. Here, we identify bone lymphatics as an essential component of the regenerative microenvironment in the mandible and long bones. We demonstrate that bone lymphatic endothelial cells (LECs) constitute a specialized endothelial population that is transcriptionally and spatially distinct from periosteal LECs. During skeletal repair, bone LECs reactivate a regenerative transcriptional programme, and promote mandibular and fracture healing. In osteonecrosis of the jaw and periodontitis, bone lymphatic-associated signalling is disrupted, identifying impaired lymphatic function as a shared feature of mandibular disease. Therapeutic activation of VEGFC-FLT4 signalling during injury or mandibular disease restores lymphangiogenesis, enhances osteogenesis, and markedly improves bone regeneration. Together, our findings advance the paradigm-shifting role of bone lymphatics positive regulators of bone regeneration and identify lymphatic activation as a promising therapeutic strategy to enhance bone regeneration in mandibular diseases.

cell biology↗

bioRxiv2026

Multi-modal Graph Integration for Biologically Interpretable Domain Identification Using Spatial Transcriptomics

Functional domain identification in spatial transciptomics transforms spatial molecular measurements into mechanistic insights into tissue physiology and pathology. However, the inherent noise and sparsity of gene expression data, along with the locality-biased design of conventional graph-based approaches, fundamentally limit the accurate identification of complex tissue domains. In this study, we propose a novel Biologically Interpretable multi-modal Graph using Spatial Transcriptomics, called BIGraph-ST, that integrates pathway activity scores and histological image features for robust spatial domain identification. BIGraph-ST represents modality-specific similarity through affinity graphs and propagates spatial topology to capture higher-order connectivity within the tissue microenvironment. Experimental results demonstrated robust performance and notable improvements across multiple gold-standard benchmark datasets, particularly in cancer tissues. Moreover, BIGraph-ST provides biologically interpretable pathway-level representations of domains, which ultimately offers a valuable tool to gain biological in-sights into complex tissue architectures. The source code will be publicly available upon acceptance.

bioinformatics↗

bioRxiv2026

Structure of a key adhesin-exopolysaccharide interaction provides insights into matrix assembly in Vibrio cholerae biofilms

Biofilms serve as a protective mechanism for bacteria, including many pathogens. To form such communities, bacteria secrete macromolecules that form an extracellular matrix serving as a barrier against environmental threats, such as predation, antibiotics, and the host immune system. To be effective, a biofilm must anchor to foreign surfaces and retain sufficient stiffness, in the environment or in a host. However, how this matrix self-organizes to support biofilm formation remains a mystery at the molecular level. The human pathogen Vibrio cholerae produces biofilms primarily composed of an exopolysaccharide called VPS (Vibrio polysaccharide), consisting of an unusually-modified repeating tetrasaccharide core unit. VPS engages with two secreted adhesion proteins, Bap1 and RbmC, which adhere the biofilm to abiotic and biotic surfaces, and serve to strengthen the biofilm by interacting with VPS using a conserved {beta}-propeller. To pinpoint the interaction between the adhesins and purified segments of VPS, we determined the [~]1.6 [A] X-ray crystal structure of Bap1 bound to fragmented VPS and used the structure to carry out molecular dynamics simulations. The structure revealed a single binding site consisting of one tetrasaccharide unit involving an induced magnesium binding site. Unexpectedly, the tetrasaccharide adopted a bent state caused by a rotation of the glycosidic bond between the central two monosaccharide units. Using a combination of mutagenesis, light scattering, and in situ fluorescent microscopy, we demonstrate that Bap1 not only facilitates biofilm adhesion, but is also required for proper VPS organization, through the identified binding pocket. Our structure reveals for the first time the interaction between a biofilm exopolysaccharide and matrix protein, as well as insights into conformational changes of exopolysaccharide induced by this binding. Our findings provide a generalizable approach for studying the biophysical and biochemical properties of carbohydrate-dependent biofilm assembly, which may lead to new ways to treat disease caused by biofilm-forming bacterial pathogens by disrupting the exopolysaccharide-protein interactions.

microbiology↗

bioRxiv2026

Mitochondrial and protein homeostasis pathways are transcriptionally impaired in islets during type 1 diabetes pathogenesis

The decline in first-phase insulin response (FPIR) during the presymptomatic period of type 1 diabetes (T1D) is well established. In-situ functional studies with pancreas tissue slices showed that {beta}-cell loss of glucose-responsiveness was independent of T-cell infiltration into islets in recent-onset T1D cases. However, the mechanisms driving {beta}-cell dysfunction before the onset of T1D remain unclear. In pancreas tissue from donors across the natural history of T1D, we utilized an in-situ, whole-islet phenotypical and transcriptomic approach to unravel novel targets in the glucose-stimulus coupled secretion pathway that are similarly impaired in T-cell infiltrated and non-infiltrated islets. Specifically, we observed that islets from autoantibody positive (single(s) or multiple(m) AAb+) donors exhibited activation of post-transcriptional gene regulation along with reduced protein translation, processing in the endoplasmic reticulum (ER), and ER stress. Disrupted mitochondrial metabolism and bioenergetics were prominent in islets from multiple AAb+ and T1D donors with disease durations [≤]7 years. In addition, T1D islets presented reduced mitochondrial protein import, quality control, and dynamics, together with downregulated genes in insulin secretory pathways. During infiltration, these pathways remain dysregulated while immune/inflammatory transcripts were increased. These studies identified novel mechanisms of {beta}-cell dysregulation before symptomatic onset and independent of T-cell infiltration in T1D pathogenesis.

physiology↗

bioRxiv2026

Elucidating enzyme-substrate specificity through co-folding foundation model

Enzymatic catalysis relies on precise structural and chemical complementarity, yet systematically mapping enzyme-substrate interactions remains a critical bottleneck. While structure-aware methods have advanced functional annotation, their reliance on predefined binding pockets and rigid-body docking fails to capture the ligand-induced conformational changes essential for catalytic turnover. Here we introduce Boltz2ESI, an end-to-end framework that predicts enzyme-substrate interactions by leveraging structural knowledge learned by a biomolecular foundation model. Through native co-folding, the framework inherently captures active-site plasticity without requiring predefined pocket annotations. Integrating these learned biophysical priors with global evolutionary context and geometric molecular descriptors, Boltz2ESI consistently outperforms state-of-the-art sequence-based and rigid-docking approaches. Extensive validation demonstrates that the framework accurately discriminates tight sub-family specificities, enabling effective candidate prioritization for biosynthetic pathway elucidation, as demonstrated on the withanolide pathway. Ultimately, this structure-dynamic approach establishes an actionable foundation for accelerating rational biocatalyst discovery and large-scale pathway de-orphaning.

bioinformatics↗

bioRxiv2026

Novel transplantable mouse cell line model recapitulates invasive lobular breast carcinoma (ILC) phenotype and immune microenvironment.

Invasive lobular breast carcinoma (ILC) is the most common special histological subtype of breast cancer, which accounts for 10-15% of all cases. To study the phenotype characteristics, metastatic growth kinetic and immune microenvironment of ILC, we developed an orthotopically transplantable cell line model from the spontaneous mammary fat pad tumor of CDH1-PTEN dual knockout C57BL/6 mouse with Cre-loxP system, designated CPT6. CPT6 recapitulates single-file growth pattern of human ILC, with pleomorphic features and a high mitotic index. RNA sequencing together with whole exome sequencing reveals a luminal A subtype with targetable driver mutations such as Kras G12C. As a novel orthotopically transplantable ILC model in immune competent mice, CPT6 shows robust in vivo growth and metastatic rate, and has moderate immunogenicity which appears to be T-cell independent. We also profiled the immune microenvironment of CPT6, revealing a myeloid-rich environment with dominant M2-macrophage population, which is concordant with human ILC. In summary, this model recapitulates human ILC phenotype and represents a valuable preclinical platform for evaluating immunotherapy and other therapeutic strategies for invasive lobular breast carcinoma.

cancer biology↗