Biology subjects

Wei, Z.

Publications and source records attributed to Wei, Z..

At least 19 recordsLinked to original sources

bioRxiv2026

Reduced cerebral blood flow and metabolic rate trigger an adaptive mitochondrial remodeling program in a Down syndrome mouse model

Mitochondrial dysfunction is a central feature of Down syndrome (DS), yet how trisomy 21 affects brain bioenergetics across scales--from gene transcription to cellular function to whole-brain metabolism--remains poorly understood. Here, we use the Dp16(1)Yey/+ (abbreviated Dp16) DS mouse model, which carries ~115 triplicated gene orthologs of human chromosome 21, to address this major knowledge gap and to provide direct in vivo evidence of brain-wide metabolic deficit in DS. Using magnetic resonance imaging (MRI) approaches, we show that Dp16 mice have reduced cerebral blood flow and metabolic rate of oxygen, with effects more pronounced in males than in females. Despite the suppression of cerebral metabolic rate, mitochondrial respiratory capacity and ATP synthase activity are largely preserved across brain regions, with sex differences noted. Transcriptomic analyses reveal marked upregulation of mitochondrial OXPHOS genes and the convergence of biological pathways related to proteostasis and mitochondrial function across brain regions, with male cortex and female cerebellum showing the biggest change. Collectively, our data suggest that a brain-wide metabolic deficit in Dp16 mice triggers an adaptive mitochondrial remodeling program across the brain to support bioenergetic demands. This compensatory response, however, is insufficient to normalize the metabolic dysfunction in Dp16 mice. Our findings underscore the value of integrating functional assessments at the organismal and tissue levels with transcriptomic changes to gain a systems view of physiologic perturbations and adaptations in the DS brain which may influence phenotypic outcomes.

physiology↗

bioRxiv2026

How Much Can AI Virtual Cells Gain from Unified Perturbational Modeling?

Predicting how individual cells respond to perturbations is a central task in artificial intelligence virtual cell (AIVC) modeling. However, integrating perturbation datasets is challenging because they differ in measured gene sets, perturbation types and cellular contexts. Existing AIVC predictors are typically developed for specific settings, limiting their applicability across heterogeneous datasets. Here, we investigate the extent to which AIVC predictors can benefit from unified perturbational modeling (UPM), a novel paradigm designed to integrate heterogeneous Perturb-seq datasets into a single predictive model. To this end, we introduce three unification modules into a variational autoencoder (VAE)-based AIVC predictor to address these sources of heterogeneity: gene-channel alignment, shared perturbation encoding, and joint embedding of gene expression and identity. We systematically evaluate and validate the individual contributions of these modules. Building on UPM, we assemble a dataset of 110 million cells spanning over 22,000 genes, four perturbation types and 76 cellular contexts, and use 89.7 million cells from 71 contexts to train a unified AIVC predictor with more than one billion parameters. Across genetic perturbation benchmarks, UPM achieves the highest PES and gene-wise response correlation among the evaluated models. It also improves all four evaluation metrics over State on the five held-out Tahoe-100M cell lines and over Stack on the cytokine benchmark. UPM also improves combinatorial perturbation prediction relative to specialized predictors. In the HSC application, our model achieves a 40.0% improvement in accuracy on a downstream task involving the recapitulation of effects induced by the knockdown of hepatic stellate cell (HSC)-related targets. Our implementation is openly available at https://github.com/youlab2025/UPM.

bioinformatics↗

bioRxiv2026

A telomere-to-telomere genome of Holstein cattle resolves chromosome-end architecture and the genetic basis of telomere length

Telomeres safeguard chromosome-end integrity, yet population-scale dissection of telomere-length genetics has been largely restricted to humans. Here, we report a 3.01-Gb telomere-to-telomere (T2T) genome of Holstein cattle that resolves repeat-rich chromosome ends and centromeres and enables telomere-length quantification at single-chromosome-end resolution. Holstein chromosome ends show marked telomere-length asymmetry and end-specific variation in telomere variant repeats, subtelomeric tandem repeats, GC/CpG composition, and DNA methylation. Using sequencing-depth-corrected telomere-length phenotypes from 9,591 cows, teloGWAS identifies major loci encompassing RTEL1, CTC1, and TERT, while multi-tissue eQTL, sQTL, TWAS, and co-expression analyses prioritize CTC1 and implicate DNA replication, DNA repair, and stress-response networks. Together, these findings extend population-scale telomere genetics beyond humans and establish a sequence-resolved framework for chromosome-end biology and telomere-informed livestock genomics.

genomics↗

bioRxiv2026

Single-cell spatial transcriptomics of inherited retinal degeneration reveals broad and focal organization of disease responses

Inherited retinal degenerations are largely monogenic disorders characterized by progressive photoreceptor dysfunction and death. Although the transcriptomic responses to inherited degeneration are well-characterized, how these responses are spatially organized across the retinal axis has not been mapped. Here, we used seqFISH to spatially profile 173 cell identity and degeneration-associated genes with single-cell, single-transcript resolution in male wild-type and LCA5gt/gtmouse retinas at early, intermediate, and advanced disease stages. In early disease, rod injury and stress response markers were enriched in the superior retina, while rod function and identity markers were enriched in the inferior retina. Nested within the broader superior-inferior organization, a coordinated rod injury response comprising Edn2, Fgf2, Socs3, and Stat3 was organized into focal regions on the order of 100 m in width, with high-response regions directly adjacent to low-response regions. These focal regions were locally enriched for markers of Muller reactive gliosis and rod stress responses. Our single-cell spatiotemporal map reveals that transcriptional responses to degeneration vary at both the broad and focal scales, and offers a framework for investigating how local tissue properties and interactions among neighboring cells may contribute to disease progression.

neuroscience↗

bioRxiv2026

Genetic risk and inflammatory signaling converge on cell type-specific regulatory programs in type 1 diabetes

Type 1 diabetes (T1D) is a complex autoimmune disease characterized by the destruction of insulin-producing pancreatic {beta} cells. Both genetic susceptibility and epigenetic dysregulation contribute to T1D risk. Recent studies have reported {beta}-cell dysfunction before disease onset, suggesting that {beta}-cell-intrinsic mechanisms contribute to pathogenesis. However, the mechanistic link between disease-associated variants and {beta}-cell dysfunction remains poorly understood. We hypothesized that a subset of T1D-associated variants directly influences cell type-specific candidate cis-regulatory elements (cCREs) by disrupting transcription factor (TF) binding and altering gene expression. To define cell type-specific cCREs in human islets, we integrated publicly available epigenomic datasets, including bulk histone modification profiles and single-cell multiomic data from human pancreatic islets. We combined these data with T1D genome-wide association study (GWAS) variants to identify disease-associated SNPs located within enhancer regions. Using multiomic datasets from PANC-DB, we characterized cell type-specific chromatin accessibility in nondiabetic and T1D islets and identified regulatory regions that may control gene expression in {beta} cells and other islet cell types. We then applied ChromBPNet, a deep learning model that predicts base-pair-resolution chromatin accessibility from scATAC-seq data, to evaluate how specific variants may alter local regulatory activity. In parallel, we used TF footprinting to nominate TFs likely to bind these variant-containing regions. These analyses identified several T1D-associated SNPs predicted to alter chromatin accessibility at candidate TF binding sites, suggesting mechanisms by which noncoding variants may contribute to {beta}-cell dysfunction and T1D susceptibility. Our findings link T1D-associated variants to cell type-specific enhancer activity and regulatory pathways and provide candidates for future mechanistic studies.

genomics↗

bioRxiv2026

FedEdgeR: federated and privacy-preserving edgeR for differential gene expression analysis

Motivation: Multi-center RNA-seq studies improve statistical power, but privacy regulations restrict patient-level data sharing. Meta-analysis methods avoid this restriction but lose power, especially under per-site imbalance. Federated learning allows sites to share only summary statistics. Among the three dominant differential expression (DE) frameworks, Flimma federates limma-voom and FedPyDESeq2 federates DESeq2. edgeR, still preferred for small-sample and high-variability designs, lacks a federated counterpart. Its iteratively reweighted least squares (IRLS) and Cox-Reid dispersion estimation make federation harder than limma-voom's single-pass fit. Results: We present FedEdgeR, a federated edgeR implementation protected by secure multi-party computation (SMPC), covering IRLS GLM fitting, three-level dispersion estimation, and the likelihood-ratio test. We evaluate it on four representative RNA-seq datasets: two tumor/normal cohorts, a multi-cohort anti-PD-1 immunotherapy study, and a 6-sample paired stress-test. FedEdgeR matches pooled edgeR under various metrics, such as Pearson r [≥] 0.99999 on -log10 p-values, 100% complete top-100 DE gene overlap, and F1 = 1.000 at the nominal FDR level 0.05. It consistently outperforms Fisher, Stouffer, random-effects, and RankProd meta-analysis on every tested dataset.

bioinformatics↗

bioRxiv2026

Deciphering Mechanistic Signatures in Drug-Drug Interactions with Dual Topology Graphs

Drug-drug interactions (DDIs) represent a critical challenge in drug development and clinical practice, as they can lead to severe adverse effects, including toxicity and reduced therapeutic efficacy. Deep learning methods have shown promise in large-scale, rapid DDI prediction; however, current approaches suffer from significant limitations in providing mechanistic insights into these interactions. Here, we propose DualTopoDDI, a dual-topology-enhanced interpretable deep learning model for DDI prediction. We applied DualTopoDDI to predict 9.2 billion potential interactions among approved drugs, achieving 97.99% high-confidence predictions. The model demonstrates molecular-level interpretability, identifying key substructures responsible for drug actions in terms of both atom-centric and bond-centric views. DualTopoDDI also excels in elucidating particular DDI toxicity mechanisms, which is validated by its successful explanation of the controversial cardiac toxicity in two COVID-19 drug combination regimens at the time. Evaluations across 11 benchmark datasets demonstrates that DualTopoDDI not only achieves state-of-the-art performance but also showcases robust generalizability and good interpretability. Overall, DualTopoDDI offers a powerful, interpretable tool for understanding and predicting drug-drug interactions, providing critical insights for drug safety and design.

bioinformatics↗

bioRxiv2026

Predicting Protein-RNA Binding Affinity Changes via Spatial Coupling-Aware State Space Modeling

Accurately predicting the effects of mutations on protein-RNA binding is crucial for elucidating disease mechanisms. Yet, exhaustively exploring the space of all possible variants is prohibitively expensive, motivating computational methods that can quantify mutation-induced changes in binding affinity (aka {Delta}{Delta}G) accurately and efficiently. We present iSCALE, an interpretable and generalizable deep learning method that adopts an implicit Spatial Coupling-Aware Ligand Encoding strategy to predict mutation-induced binding affinity changes. By injecting this implicit multiscale encoding scheme into a bidirectional state space modeling architecture, iSCALE learns a generalizable multiscale coupling pattern that achieves superior performances on not only the protein-RNA binding {Delta}{Delta}G, but also the protein stability {Delta}{Delta}G and protein-protein binding {Delta}{Delta}G predictions. Detailed analyses demonstrate that the model attention scores align well with structural characteristics. In addition, iSCALE shows good discriminative ability when predicting close samples such as complexes of same mutation but with different ligands or the same complex but with different mutation sites. In summary, iSCALE serves as an effective in silico tool for large-scale protein-RNA binding {Delta}{Delta}G prediction, which pushes the border of understanding in mutation-induced pathological outcomes.

bioinformatics↗

bioRxiv2026

Covariance Nonstationarity is Evident in Spatial Transcriptomics and Provides a New Categorization of Spatially Varying Genes

Gaussian process models underlie many spatial transcriptomics tools but typically assume stationary covariance. Covariance non-stationarity has long been recognized in spatial statistics as an important feature of spatial data, yet it has received little attention in spatial transcriptomics. We show that this omission is consequential: covariance non-stationarity is substantially evident across spatial transcriptomic datasets and alters the characterization of spatially varying genes. While typically ignored, non-stationarity of spatial covariance in gene expression may correspond to tissue heterogeneity or cell aggregates. Across 12 Visium datasets, we use approximate Bayes factors from R-INLA to compare stationary and non-stationary Mat'ern covariance functions. Evidence for covariance non-stationarity appears in 3% to 50% of genes across tissue samples. We find that gene sets associated with immune, cytokine, and other effector functions are enriched among genes favoring non-stationary spatial covariance. Covariance stationarity is therefore not a benign technical simplification in spatial transcriptomics; it is frequently violated, the violation is biologically structured, and it changes the definition and classification of spatially varying genes.

bioinformatics↗

bioRxiv2026

onsite: An Integrated Framework for Phosphosite Localization and False Localization Rate Estimation

With the rapid development of mass spectrometry-based proteomics, the volume of phosphoproteomic data has increased substantially. However, accurate localization of phosphorylation sites and standardized statistical validation remain critical analytical bottlenecks. To address the lack of standardized cross-algorithm evaluation, we introduce onsite, a unified and open-source Python framework. onsite integrates an alanine-decoy strategy to estimate the false localization rate (FLR) across three algorithms: AScore, PhosphoRS, and pyLucXor. This modular architecture efficiently processes large-scale datasets and enables global FLR calculation. Benchmarking on the standard synthetic phosphopeptide dataset PXD000138 highlighted distinct inter-algorithmic variations. Using the same 5% global FLR threshold, pyLucXor localized the most target sites (28,353). It also reached a high accuracy (91.22%) against the known ground truth, resulting in the largest number of correctly localized sites (25,865). Reanalysis of the highly fractionated, large-scale PXD012255 dataset further demonstrated that native integration of onsite into the quantms pipeline enables scalable processing and provides a standardized framework for FLR control in large-scale phosphoproteomics. Graphical Abstract O_FIG O_LINKSMALLFIG WIDTH=200 HEIGHT=64 SRC="FIGDIR/small/737157v1_ufig1.gif" ALT="Figure 1"> View larger version (14K): org.highwire.dtl.DTLVardef@e4c85dorg.highwire.dtl.DTLVardef@1e8464org.highwire.dtl.DTLVardef@185cea1org.highwire.dtl.DTLVardef@1c0d1bc_HPS_FORMAT_FIGEXP M_FIG C_FIG

bioinformatics↗

bioRxiv2026

CellPilot: an agentic framework that pilots small language models through autonomous single-cell annotation

Large language models can annotate cell types from marker gene lists, but they typically operate after preprocessing and clustering are complete, treating annotation as a terminal labeling step rather than controlling the analytical decisions that produce the evidence for cell identity. We present CellPilot, an agentic framework that guides a locally deployable small language model through the full single-cell analysis workflow, from raw count matrices to cluster-level annotation. CellPilot combines standard single-cell analysis tools with structured workflow control and observation-guided reasoning, allowing the model to plan analyses, execute tools, inspect intermediate results and revise decisions within a traceable session. On GTEx, structured workflow orchestration raised the same 8B model from 0.39 in a prompt-only setting to 0.89, closing most of the gap to GPT-4o (0.92) within the same framework; the framework gain was substantially larger for the smaller backbone across datasets (+0.35 versus +0.19). Across GTEx, Tabula Sapiens, and Mouse Cell Atlas, CellPilot achieves cluster-level annotation accuracies of 0.891, 0.750, and 0.773, outperforming representative reference-based, marker-based, and LLM-based methods. CellPilot confidence scores were associated with annotation correctness and supported post hoc filtering, while complete execution traces were retained for each analysis. These results suggest that structured workflow orchestration can be a critical determinant of performance in multi-step single-cell analysis, enabling locally deployable small language models to approach larger proprietary models while preserving transparency and practical usability.

bioinformatics↗

bioRxiv2026

An allosteric pocket in KV1.3 defines a distinct chemical space for immunomodulator design

KV1.3 is a validated autoimmune drug target, yet all disclosed inhibitors converge on a narrow set of conserved binding sites, suffering from poor selectivity or clinical failure. Whether KV1.3 harbors a distinctive, druggable pocket amenable to selective targeting has remained unknown. Here we report MPiN, a highly selective, state-dependent KV1.3 inhibitor with in vivo efficacy in a mouse psoriasis model, which engages a previously uncharacterized extracellular allosteric pocket framed by the PP1-PP2 turret loops, the pore helix and the outer S5/S6 helices, acting through a bidirectional pore-to-sensor coupling that simultaneously constricts the selectivity filter and facilitates the voltage sensor toward activation. Unexpectedly, despite the extensive structural conservation of this pocket across KV1 paralogs, subtype selectivity is dictated by the peripheral residues G427 and H451, which define pocket geometry without directly contacting the ligand, thereby establishing a geometry-driven "non-contact selectivity" mechanism. By opening an unrecognized, structurally distinct chemical space on KV1.3 and redefining how selectivity is achieved within a conserved channel family, this work lays the structural and conceptual foundation for rational, structure-guided design of next-generation KV1.3 immunomodulators.

biochemistry↗

bioRxiv2026

Improving Generalizability in Whole-Cell Antibiotic Discovery Through Active Learning

Machine learning (ML) has accelerated molecular discovery, yet training models to generalize to out-of-distribution (OOD) chemical spaces remains fundamentally constrained by the high cost of experimental validation. In antibiotic discovery, where whole-cell phenotypic high throughput screening (HTS) is resource-intensive, iterative ML-guided compound selection - or Active Learning (AL) - offers a pathway to efficiently navigate available chemical spaces. However, the algorithmic tradeoffs between prioritizing compound novelty (exploration), predicted bioactivity (exploitation), and their impact on OOD generalizability remain unresolved for noisy, whole-cell biological systems. In this work, we systematically evaluate three AL strategies for whole-cell bacterial bioactivity and benchmark their effects on model accuracy, hit rate, and OOD performance. Using retrospective simulations on Mycobacterium tuberculosis HTS data, we identify an optimal AL strategy that balances predicted hit/non-hit novelty with overall hit rate. We then integrate the strategy in a closed-loop Borrelia burgdorferi antibiotic discovery HTS campaign. The AL-guided approach successfully increased the experimental screening hit rate five-fold (from a 0.2% rate within investigator-selected plates to 1.0%). Further, when the trained model was applied in prospective in silico selection of highly diverse compounds across multiple bacterial species, the AL-trained whole-cell inhibition predictor demonstrates 53-fold enrichment over investigator-directed screening (11.0% experimental validation of predicted hits). Of these, 100% demonstrated the intended narrow spectrum activity for Borrelia burgdorferi. These results demonstrate that calibrated AL strategies can overcome data acquisition bottlenecks and train generalizable property predictors able to extrapolate to OOD molecules.

bioinformatics↗

bioRxiv2026

Multisensory integration of a host metabolite coordinates bacterial predation of macrophages

The crisis of antimicrobial resistance demands new strategies that circumvent conventional killing. Here, we show that Yersinia pseudotuberculosis intercepts host mesifurane, a mimic of quorum-sensing signal AI-2. Rather than acting through a single receptor, mesifurane is directly and independently sensed by three receptor classes, YcrA, NarX, and CusR, that converge on a "Sense, Arm, Recruit" program. YcrA drives chemotaxis toward host cells; NarX-NarP licenses T6SS-mediated killing; and CusR amplifies AI-2 production to recruit more bacteria. YcrA is conserved across Gram-negative pathogens, creating a therapeutic vulnerability. Suramin sodium, a repurposed drug blocking YcrA, disarms this circuit without inhibiting growth. In murine sepsis, suramin sodium conferred 80-90% survival against WHO critical-priority pathogens, including carbapenem-resistant A. baumannii, E. coli, and P. aeruginosa, even when delayed 8 hours post-infection, without selecting for resistance. These findings establish that disrupting sensory perception, rather than viability, provides an anti-infective strategy bypassing resistance while imposing minimal selective pressure.

microbiology↗

bioRxiv2026

Polarity remodeling of ventricular cardiomyocyte precursors during heart morphogenesis is required for ventricular chamber formation

Vertebrate heart development requires precise regulation of morphogenetic movements and dysregulation causes congenital heart defects. However, cell polarity remodeling during early heart tube formation remains underexplored. Here, we show that zebrafish ventricular cardiomyocyte precursors (vCPCs) undergo striking polarity reorientation as they transition from a single-layered epithelium to a transient double-layered configuration by involution. During this process, vCPCs remove inner adherens components and re-establish polarity toward the ECM-attached outer surface, enabling proper directional extension to form the ventricle. We identify the junctional scaffold protein Afadin a (Afdna) as a critical regulator. Afdna localizes to the outer and lateral membrane to restrict Rap1 activity, facilitating Podocalyxin translocation and adhesion disassembly at the inner interface between vCPC double layers. afdna mutants exhibit a multilayered ventricle with impaired blood flow, phenocopying human congenital ventricular obliteration. Our findings uncover a polarity-based mechanism ensuring ventricular chamber formation and provide insight into ventricular obliteration.

developmental biology↗

bioRxiv2026

The unconventional kinesin Kif26a is required for the guidance of major axon tracts in the developing mouse brain

Kif26a and Kif26b encode a family of unconventional kinesins with emerging roles in neurodevelopment, and mutations in both genes have been implicated in a spectrum of neurodevelopmental disorders. The precise mechanisms by which the Kif26 family orchestrates mammalian brain development, however, remain unclear. In this study, we show that Kif26a and Kif26b are expressed in distinct regions of the developing mouse brain. Using a new allelic series in mice, we demonstrate that Kif26a is required for the guidance of multiple forebrain axon tracts. This requirement is direct and cell autonomous, as cell proliferation, survival, and cortical layering are unaffected in the Kif26a mutants. These guidance defects closely resemble those reported for the Fzd3-Celsr3-Dystroglycan pathway, suggesting that Kif26a may participate within this conserved signaling axis to steer growing axons.

neuroscience↗

bioRxiv2026

Structure-guided reinterpretation of a disease-associated CWH43 residue-533 truncation identifies internal disruption of a conserved C-terminal module in idiopathic normal pressure hydrocephalus

Idiopathic normal pressure hydrocephalus (iNPH) is a chronic neurological disorder characterized by ventricular enlargement together with gait impairment, cognitive decline, and urinary dysfunction. Disease-associated CWH43 variants have emerged as genetic clues in iNPH, but the molecular consequences of truncating variants remain poorly defined. Here, we used structure-guided analysis to reinterpret the disease-associated human CWH43 variant c.1596del, annotated as p.Leu533Ter and referred to here as the residue-533 truncation. AlphaFold-based modeling of the full-length 699-amino-acid protein supported a conserved C-terminal D3 module spanning residues 407-699. Within this framework, the residue-533 truncation is best interpreted as an internal interruption of D3 rather than deletion of a poorly constrained terminal tail. Pocket analysis identified a dominant candidate surface site within D3 containing 18 pocket-lining residues, 55.6% of which (10/18) are removed by truncation. Independent cavity mapping further showed that the affected region is better understood as a broader cavity-bearing architecture rather than a single local pocket. Cross-boundary contact analysis, multi-model reproducibility, and representative local geometry supported structural coupling between retained and deleted D3 segments, indicating that truncation is predicted to cause retained-side decoupling in addition to direct residue loss. Together, these findings support a model in which the disease-associated residue-533 truncation disrupts a structurally integrated C-terminal module and its associated cavity-bearing architecture. This framework refines interpretation of CWH43-associated iNPH variants and provides a mechanistic bridge between human genetic association and architecture-level loss of function.

genetics↗

bioRxiv2026

Coordinated phase separation and phase transition underlie synaptic ribbon condensate formation and plasticity

Mesoscale molecular apparatuses with diverse functions are widespread in cells. How mesoscale cellular apparatuses are formed and regulated is poorly understood. The synaptic ribbon in sensory neurons, which can dynamically adjust its size and shape in response to light or sound, is a mesoscale molecular assembly. In this study, we demonstrate that RIBEYE, the backbone protein of synaptic ribbons, via combined ordered and disordered interactions, forms micron-sized dynamic molecular assemblies. Integrated cryo-electron microscopy and tomography studies revealed that the N-terminal SAM domain and the C-terminal B domain of RIBEYE form distinct nanoscale filaments and further assemble into ribbon-like architectures via phase transition. Unexpectedly, the N-terminal intrinsically disordered region of RIBEYE softens the solid mesoscale ribbon structures via phase separation. The mesoscale size and shape of the RIBEYE condensate are bidirectionally tuned by physiological modulators such as Piccolino and CtBP1, linking synaptic ribbon structural plasticity to functional adaptability. Thus, our study demonstrates that combined phase transition and phase separation of nanoscale protein oligomers can build mesoscale functional cellular apparatuses.

biochemistry↗