Biology subjects

Chen, Z.

Publications and source records attributed to Chen, Z..

At least 91 records · Page 5Linked to original sources

bioRxiv2026

Prenatal e-cigarette exposure induces region- and cell type-specific metabolic and developmental reprogramming

Prenatal e-cigarette exposure (PeCE) is increasingly prevalent and has been associated with adverse neurodevelopmental outcomes, yet how maternal vaping perturbs early brain development remains poorly understood. We integrated spatial transcriptomics, snRNA-seq and lipidomics to define neonatal rat brain responses to PeCE in rats at regional and cellular resolution. PeCE induced pronounced spatial heterogeneity in vulnerability, with the striatum exhibiting the strongest developmental transcriptional disruption. PeCE disrupts lipid metabolic homeostasis and induced molecular signatures suggestive of enhance Ca2+ signaling, dopaminergic responsiveness and region-specific synaptic stress programs. In the striatum, PeCE induced transcriptional programs consistent with altered lipid utilization, silent synapse-like molecular features, suppressed dendritic spine development and delayed D1-medium spiny neuron maturation. PeCE-sensitive genes were enriched in human autism spectrum disorders and neurodevelopment risk loci. Together, these findings identify disrupted metabolic and developmental reprogramming as a central feature of neonatal brain vulnerability to maternal vaping and provide a mechanistic framework linking PeCE to neurodevelopmental disease risk.

neuroscience↗

bioRxiv2026

Electrolytic-Microbubble Dynamics Delineate Safety Thresholds During Intracortical Microstimulation with Flexible Neural Interfaces

Intracortical microstimulation (ICMS) with ultraflexible neural electrodes enables low-threshold, chronically stable, and high-resolution modulation of neural circuits, providing a promising strategy for sensory restoration and closed-loop neuromodulation. However, the microscopic mechanisms delineating its safe and effective current range remain unclear. Here, we combine intravital two-photon (2P) imaging and electrophysiology in awake mice to examine the current-dependent neurovascular outcomes of charge-balanced stimulation via ultraflexible arrays. We observed gas bubbles formed along the electrode during ICMS, with bubble size increasing quadratically with current amplitude, consistent with a Faradaic bubble-growth model. Intravital 2P imaging reveals that at low-to-moderate currents (20-40 {micro}A), vascular leakage is small, spatially confined, and largely reversible, whereas higher currents ([≥]60 {micro}A) induce a sharp transition to extensive, field-dominated extravasation and secondary vessel disruption. This transition coincides with immediate, stimulus-locked motor responses and the onset of electrode degradation. Multiphysics simulations reproduce the observed nonlinear leakage-current relationship by incorporating gas bubble-induced electric field redistribution and voltage-dependent vessel wall permeability. The model indicates that gas bubbles act as local electric-field modulators, concentrating suprathreshold fields near the bubble boundary at lower currents while shielding more distant vessel segments; at higher currents, this confinement breaks down and the system enters a field-dominated damage regime. Collectively, these findings define a mechanistically informed safety window for ICMS with flexible neural interfaces and identify bubble-assisted vascular permeabilization as a key failure mode at high currents, crucial for the design of future bidirectional brain-computer interfaces and high-precision neuroprosthetic protocols.

neuroscience↗

bioRxiv2026

Seeing the world in a grain of sand: Fine global and local representations in foveal and parafoveal vision

Foveal vision enables primates to perceive small objects with remarkable precision by resolving both fine-global form and fine-local detail. Yet current coarse-to-fine and local-to-global frameworks, derived largely from parafoveal recordings and large-object representations, posit that V1 encodes local detail whereas IT represents global form, leaving the neuronal basis of foveal object representation unresolved. Here we show that fine-global form and fine-local detail are encoded parallelly as early as V1 and preserved across V2, V4 and IT at both foveal (1{degrees}-2{degrees}) and parafoveal (2{degrees}-6{degrees}) retinal eccentricities. Coarse-global features dominate parafoveal vision and higher cortical areas. Response latencies revealed a rapid processing sequence, with 2-6 ms separating coarse-global, fine-local, and fine-global processing within parafoveal V1, V2 and V4. These findings establish a cortical neuronal framework to resolve fine-global form of small objects in foveal vision along primate ventral visual stream, supporting fine-scale perception during reading and fine object recognition. Highlights O_LIFoveal two-photon imaging across V1, V2, V4, and IT C_LIO_LIGlobal and local processing examined across eccentricity C_LIO_LIFine and coarse global representations emerge as early as V1 C_LIO_LIFine global and local representations are preserved from V1 up to IT C_LIO_LITightly timed coarse-to-fine-to-fine processing within each cortex C_LIO_LIFast coarse-global dominates parafoveal and higher cortical areas C_LI

neuroscience↗

bioRxiv2026

Healing cascades and infections in wounds monitored using a wearable sensor of gaseous flux

Capabilities for quantitative monitoring of chronic wounds remain an unmet clinical need, as existing diagnostic approaches rely on semiquantitative evaluation of symptoms that lack sensitivity especially during early stages of infection. Here we present a scheme for tracking wound physiology that leverages a miniature, wireless skin-interfaced device for non-contact, transient measurements of the flux of volatile organic compounds (VOCs) and water vapor from the wound microenvironment. Unlike emerging smart bandage platforms that rely on physical contact with the fragile wound bed to interrogate liquid-phase biomarkers, this strategy uses an engineered microclimate and suspended suite of sensors to measure the diffusive transport of wound-derived gases across the wound surface but separated from it. The result enables quantitative evaluation of metabolic activity and healing progression without perturbing the healing tissues. In biofilm growth models of Staphylococcus aureus, measurements demonstrate that trends in VOC flux correlate strongly with bacterial growth kinetics and precede any visible biofilm formation. Longitudinal monitoring in infected murine wound healing models shows that concurrent measurements of water vapor and VOC flux provide complementary physiological insights, capturing both the trajectory of barrier restoration and the dynamics of bacterial burden. The findings establish this non-contact sensing scheme as a distinct and clinically translatable paradigm for wound monitoring, with broad implications for non-invasive surveillance of disease states in which tissue metabolic activity and skin barrier integrity serve as actionable physiological readouts. Significance StatementLimited capabilities in continuous, quantitative assessment of a wound make early diagnosis and effective management challenging, particularly in cases of infection that rapidly progress before symptoms appear. Non-contact approaches for wound monitoring that preserve fragile tissue can transform wound care. In this context, gaseous flux from the wound bed provides an integrative measure of microbial activity and barrier restoration. This study establishes a wearable sensing platform that quantifies these fluxes in real time, enabling early infection detection and temporal tracking of wound healing. These results highlight a path toward personalized treatment strategies and reduced reliance on episodic clinical evaluation.

bioengineering↗

bioRxiv2026

Fine-mapping candidate neuropsychiatric regulatory variants using cell type-aware comparative genomics

AbstractMeasures of nucleotide sequence conservation across species are useful for identifying functional genomic loci, but can fail when regulatory function is maintained, often in a cell type-specific manner, even when sequence is not. We introduce CTACIT, the Cell Type-Aware Conservation Inference Toolkit, to identify trait-associated regulatory variants. CTACIT integrates sequence conservation scores with cell type-specific open chromatin data collected from a few mammalian species to impute function for hundreds more. Applying CTACIT to neuropsychiatric trait loci identifies higher heritability enrichment and more fine-mapped variants than nucleotide conservation and human chromatin data alone. Our in vivo reporter assays validate predictions for enhancers with risk variants near the DRD2 schizophrenia risk locus. By integrating genome conservation and multi-species open chromatin data, CTACIT prioritizes variants within regions of conserved regulatory function for in vivo characterization and addresses a major challenge in translating disease associations to mechanistic understanding. One-Sentence SummaryRegulatory functions conserved across mammals underlying caudate cell type evolution reveal functions of human risk variants.

genomics↗

bioRxiv2026

Ventricular Expansion Couples Hyperosmotic Stress to Thirst

Thirst is thought to arise from osmotic signals detected by circumventricular organs, yet how hyperosmotic stress activates mechanosensitive channels remains unclear because cell shrinkage should reduce membrane tension. Here we identify a brain-scale mechanical signal that couples systemic osmotic stress to drinking. In mice, hypertonic saline rapidly increased serum and cerebrospinal fluid osmolality, expanded the lateral ventricles, and promoted water intake. Relieving or blocking ventricular deformation attenuated drinking without eliminating osmotic gradients. Hyperosmotic challenge also produced localized deformation of periventricular cells, where spatial transcriptomics revealed candidate mechanosensitive channels, including Tmem63b and Piezo1. Thus, ventricular expansion provides a mechanical component of osmotic thirst, uncovering an osmo-mechanical layer of interoceptive regulation.

neuroscience↗

bioRxiv2026

Identifying Ecological Restoration Priority Areas through a Function-Structure-Dynamic Framework: Integrating Ecological Security Patterns and Future Land-Use Simulation in the Anhui Section of the Yangtze River Basin

Rapid urbanization has intensified ecological degradation and threatened ecosystem sustainability in the Yangtze River Basin. This study proposes a Function-Structure-Dynamic (FSD) framework integrating ecological protection importance assessment, ecological security pattern (ESP) construction, and future land-use simulation to identify ecological restoration priorities in the Anhui section of the Yangtze River Basin. Ecologically important areas were identified by coupling ecosystem services and ecological vulnerability. MSPA, landscape connectivity analysis, resistance surface modeling, and circuit theory were applied to construct the ESP, while the PLUS model simulated land-use change and ecological risks in 2032. Results identified 37 ecological sources covering 6,901.39 km{superscript 2} and 84 ecological corridors with a total length of 916.80 km. Ecological pinch points, barrier areas, and 17 ecological warning areas were further delineated. Based on these findings, four ecological restoration zones were proposed with differentiated management strategies. The FSD framework effectively integrates ecological functions, spatial structure, and future dynamics, providing scientific support for ecological restoration and sustainable territorial spatial planning in rapidly urbanizing regions.

ecology↗

bioRxiv2026

Physics-Driven Zero-Shot Reconstruction of Isotropic 3D Fluorescence Microscopy under Undersampled Acquisition

Three-dimensional (3D) imaging represents the development of next generation of fluorescence microscopy. However, routine axial down-sampling makes isotropic resolution unrealistic. Here, we propose DeepUI, a physical zero-shot framework designed to achieve isotropic 3D fluorescence images from a low axial sampling rate. DeepUI fully leverages the intrinsic characteristics of 3D images through physics-guided degradation, which incorporates spatial-frequency joint learning to generate a scaled optical transfer function, combined with noise degradation and an up-sampling branch. Typically requiring just 5 minutes for training and 0.5 minutes for high-throughput and fast prediction, we demonstrate the superior performance of DeepUI to get isotropic results, and the exclusivity to axial down-sampling conditions, even in more challenging conditions, including defocused background, noise, and resolution blur.

bioinformatics↗

bioRxiv2026

Lineage Memory and Convergent Adaptation Dictate the Single-Cell and Spatial Architecture of Brain Malignancies

How diverse cancer lineages navigate the draconian central nervous system environment--whether constrained by ancestral ontogeny or driven by convergent adaptation--remains a fundamental biological paradox. To decode this, we integrated extensive in-house and public resources to construct the largest single-cell atlas of primary and secondary brain malignancies (>550,000 cells). We demonstrate that malignant cells balance strict lineage imprinting with shared brain-adaptive programs, powerfully converging upon a dominant pan-cancer mesenchymal (MES)-like state alongside a SYT1+ pioneer subpopulation exploiting neuronal mimicry. Concurrently, the microenvironment undergoes lineage-constrained divergence. The myeloid landscape shifts from a resident-dominated architecture in gliomas to extensive blood-borne infiltration in metastases. Lymphoid responses parallel this, crowning laryngocarcinoma as an ultra-hot subtype and revealing that massive T cell influx in highly infiltrated niches is paradoxically driven into terminal exhaustion, contrasting with severe immune exclusion in cold tumors. Furthermore, stromal-vascular adaptation couples pro-angiogenic tip-endothelial activation with a dynamic opposition between structural extracellular matrix (ECM) deposition and focal proteolytic degradation. Anchored by the pan-cancer MES adaptation, we leveraged subcellular spatial transcriptomics to redefine the microenvironment around four MES-like architectural niches, mapping an evolutionary trajectory from perivascular entry to immunosuppressive stromal remodeling. Finally, projecting these architectures onto independent clinical cohort establishes the MES-S2 proliferative niche as the primary driver of severe clinical deterioration across both transcriptomic and proteomic dimensions.

Cancer Biology↗

bioRxiv2026

Epigenetic Patterns of Xylanibacter ruminicola in Bovine Rumen Across Seasons and Pregnancy

Ruminants obtain nutrients through the microbial fermentation of plant material in the rumen. Xylanibacter ruminicola is a highly abundant bacterial species in the rumen. During fermentation, X. ruminicola utilizes diverse carbohydrates from plant materials to synthesize propionate, a volatile fatty acid providing energy to ruminants. However, variation in pasture quality (e.g, nutrient and fibre content) across seasons and host pregnancy status can alter the rumen microenvironment, potentially affecting microbial activity. Bacteria in culture display distinct methylation (a reversible epigenetic modification capable of gene regulation) changes in response to the growth environment. We hypothesized that the changes to the rumen environment would affect the DNA methylation patterns within the X. ruminicola genome. Rumen fluid from 37 female Brahman cattle (17 pregnant) were sampled across four seasons. DNA methylation profiles (N6-methyladenine, N4-methylcytosine, and 5-methylcytosine) of X. ruminicola across seasons (varying pasture quality) and pregnancy statuses characterized using Oxford Nanopore sequencing. After correcting for relative abundance, DNA methylation levels within the coding DNA sequences of several X. ruminicola genes differed between seasons and pregnancy status. Most of these genes were classified as ExbD/TolR family proteins and related to the protein transport process. Our study demonstrates that the DNA methylation profiles of rumen X. ruminicola genes vary with host environment factors. These results provide insight into the role of bacterial DNA methylation in mediating interactions between bacteria and their environments. Lay SummaryRuminants rely on rumen microbes to convert plants into nutrients. Xylanibacter ruminicola is a bacterial species in the rumen that produces nutrients for ruminants during plant fermentation. Changes in pasture quality and host pregnancy status can influence the activity of X. ruminicola, as reflected in the DNA methylation profile across its genome. DNA methylation is a reversible DNA modification that can affect gene activity and help bacterial adaptation to changing environments. Rumen fluid from 37 female Brahman cattle (17 pregnant) across four seasons were used to evaluate the pasture quality and host pregnancy effects on the DNA methylation profile of X. ruminicola. The relative abundances of X. ruminicola were influenced by pasture quality, but not by host pregnancy. However, host pregnancy status and changes in pasture quality influenced the DNA methylation signatures of X. ruminicola. Genes with DNA methylation changes were associated with the protein transport process. These findings suggest that the DNA methylation profiles of rumen X. ruminicola vary with host environment factors. Teaser TextThis study demonstrates that the DNA methylation profiles of rumen Xylanibacter ruminicola vary with host environment factors. These findings provide insight into the role of bacterial DNA methylation in mediating interactions between rumen bacteria and their environment.

microbiology↗

bioRxiv2026

Anthocyanin-associated cellular programs underlying terroir variation in Cabernet Sauvignon grape berry revealed by SEED-based deconvolution

Plant tissues consist of diverse cell populations that collectively contribute to development, metabolism, environmental responses, and phenotype formation. Although single-cell and single-nucleus RNA sequencing have greatly advanced the study of plant cellular heterogeneity, their application to large sample cohorts remains limited by cost, technical complexity, tissue dissociation constraints, and throughput. In contrast, bulk RNA-seq datasets have accumulated extensively across plant species, tissues, developmental stages, and environmental conditions, yet the celltype-level information embedded in these datasets remains difficult to resolve because plant-oriented deconvolution frameworks are still lacking. Existing deconvolution methods have largely been developed in mammalian systems and have not been systematically optimized for plant transcriptomic features, leaving their applicability under plant-specific constraints unclear. Here, we present SEED, an adaptive deconvolution framework optimized for plant transcriptomic data. SEED integrates candidate reference-template construction with seven deconvolution strategies and automatically identifies an optimal combination for a given dataset. In grapevine simulated benchmarking, SEED showed its clearest advantage under low-replication conditions and remained broadly competitive, rather than uniformly dominant, when larger pseudo-bulk sample sizes were evaluated. SEED further performed robustly in public Arabidopsis thaliana and Nicotiana tabacum datasets. Finally, we applied SEED to bulk RNA-seq data generated in this study from Vitis vinifera cv. Cabernet Sauvignon berries collected from Yinchuan and Yantai, identifying terroir-associated cell subtypes and coordinated celltype interaction patterns. Together, these results establish SEED as a practical framework for plant transcriptome deconvolution and provide a new tool for dissecting cellular heterogeneity associated with environmental adaptation and phenotype formation in plants.

bioinformatics↗

bioRxiv2026

RISOP, a Reference-Assisted Approach for Enhanced Identification of Oxidized Phospholipids

Oxidized phospholipids (OxPLs) play critical roles in inflammation, ferroptosis, and other oxidative stress-associated processes, yet their systematic characterization in biological systems remains a major analytical challenge owing to their low abundance and vast structural diversity. Here we report RISOP (Reference-Assisted Identification of Sample-specific Oxidized Phospholipids), an untargeted LC-MS/MS workflow that leverages chemically enriched OxPL reference pools to substantially improve OxPL annotation. Reference pools encompassing diverse oxidative modifications and a wide abundance range were experimentally generated using Fenton reaction and H2O2 treatment, providing broad coverage of OxPLs. Reference-assisted integrative analysis yielded a more than twofold increase in identified OxPL species in biological samples under elevated oxidative stress, as demonstrated in ML210-treated cells, a model of ferroptotic stress. We further show that the widely used BODIPY C11 lipid peroxidation probe captures cellular oxidative burden in only a subset of OxPLs identified by RISOP, highlighting the importance of untargeted, comprehensive OxPL profiling. Overall, RISOP provides a versatile platform readily applicable to other classes of oxidized complex lipids for comprehensive characterization in physiological and pathological contexts. Abstract Graphic O_FIG O_LINKSMALLFIG WIDTH=200 HEIGHT=105 SRC="FIGDIR/small/728773v1_ufig1.gif" ALT="Figure 1"> View larger version (24K): org.highwire.dtl.DTLVardef@482111org.highwire.dtl.DTLVardef@173126corg.highwire.dtl.DTLVardef@1be7238org.highwire.dtl.DTLVardef@3cf7f2_HPS_FORMAT_FIGEXP M_FIG C_FIG

systems biology↗

bioRxiv2026

Instant Prior-Free Resolution Enhancement for Cross-Modality Microscopy

The resolving power of optical microscopy is fundamentally constrained by the diffraction of light, limiting our ability to visualize subcellular structures. Computational methods, particularly deconvolution, can restore blurred images but critically depend on an accurate point spread function (PSF), whose estimation is often impractical and error-prone, leading to artifacts. Here, we introduce Nonlinear Fourier Re-weighting (NFR), a rapid algorithm that operates without any prior knowledge of the imaging system, achieving deconvolution-like effects through a single logarithmic mapping of the images Fourier spectrum. This non-iterative process re-balances spatial frequency components to computationally reverse the effects of optical blurring. We demonstrate that NFR robustly enhances resolution beyond the Sparrow limit and recovers authentic structural details. NFR excels where traditional methods fail, remaining effective in the presence of severe optical aberrations and high noise. Furthermore, NFR synergistically improves the output of super-resolution modalities like structured illumination microscopy (SIM), and its near-instantaneous processing enables real-time enhancement of dynamic biological processes, such as in vivo multi-photon microscopic imaging deep within scattering tissue. By decoupling high-fidelity image restoration from system modeling, NFR offers a powerful, accessible, and universally applicable tool for improving image quality across diverse microscopic techniques, facilitating the analysis of large datasets and the discovery of previously obscured biological phenomena.

bioengineering↗

bioRxiv2026

Epigenome Alterations and 3D Chromatin Architecture Remodeling in Inflammatory Macrophage Activation under Diabetic Conditions

Aberrant monocyte/macrophage activation in diabetes drives chronic inflammation and complications. While epigenetic mechanisms are implicated, the role of 3D-chromatin reorganization remains unclear. Using integrated multi-Omics, we profiled gene expression and 3D-chromatin architecture in human CD14+ monocyte-differentiated macrophages treated with high glucose + TNF- (HT) mimicking the diabetic milieu. HT induced inflammatory programs resembling those in diabetes, and dynamically altered chromatin accessibility, enhancer-promoter loops, transcriptionally active/inactive (A/B) compartments, and topologically associated domains. Inflammatory genes exhibited increased, whereas cell-cycle and metabolic genes showed reduced chromatin accessibility and enhancer-promoter interactions. These architectural changes facilitated the convergence of lineage-specific and signal-dependent transcription factors at enhancer-promoter loops, forming network hubs orchestrating pro-inflammatory responses. Similar enhancer-promoter interactions were observed in monocytes from individuals with diabetes. Perturbing chromatin interactions via CRISPR-interference targeting HT-induced enhancers suppressed inflammatory gene expression. Results from this first comprehensive enhancer connectome maps in macrophages under diabetic conditions reveal disease-associated rewiring of the 3D epigenome, highlighting epigenetic mechanisms as potential therapeutic targets. TeaserDiabetic conditions rewire 3D chromatin organization and enhancer-promoter interactions in human macrophages, promoting chronic inflammation.

genomics↗

bioRxiv2026

Fully Modified SpyCas9 Guide RNAs Enable Robust Genome Editing In Cells and In Vivo

Precision engineering of CRISPR/Cas9 components has advanced genome editing toward therapeutic applications. Completely chemically stabilized guide RNAs (gRNAs) have the potential to improve in vivo editing efficacy while enabling greater flexibility in delivery strategies. However, previous generations of fully modified guides have been associated with reduced Cas9 activity. Here, we employed an iterative, structure-guided optimization strategy to systematically introduce chemical modifications at each position of SpyCas9 gRNAs. Extending beyond commonly used nucleotide modifications, we incorporated 2-amino-RNA, 4-thio-RNA, and extended nucleic acid (exNA) to generate gRNA designs in which 90-100% of the nucleotides are sugar- or backbone-modified. Although certain modification patterns exhibit sequence-dependent variability, we have established a growing repertoire of guides that consistently maintain or enhance editing efficacy when applied both in vitro and in vivo. Collectively, our heavily and fully modified gRNAs hold potential for applications in nuclease editing, base editing, and other genome editing tools. Graphical abstract O_FIG O_LINKSMALLFIG WIDTH=200 HEIGHT=132 SRC="FIGDIR/small/725424v1_ufig1.gif" ALT="Figure 1"> View larger version (23K): org.highwire.dtl.DTLVardef@1118f08org.highwire.dtl.DTLVardef@1c57069org.highwire.dtl.DTLVardef@1572148org.highwire.dtl.DTLVardef@14a111e_HPS_FORMAT_FIGEXP M_FIG C_FIG

biochemistry↗

bioRxiv2026

Comparative profiling of carnitine palmitoyltransferase 1 isoforms reveals vincamine as a selective carnitine palmitoyltransferase 1b inhibitor

Carnitine palmitoyltransferase 1 (CPT1) catalyzes the rate-limiting step of fatty acid oxidation and has emerged as a therapeutic target for metabolic diseases and cancer. CPT1 exists in three isoforms, CPT1a, CPT1b, and CPT1c, with distinct tissue distributions and enzymatic properties; however, limitations of previous platforms enabling parallel isoform comparison have undermined efforts to identify selective inhibitors that could minimize off-target effects. Here, we describe a DTNB-based enzyme activity assay adapted for high-throughput screening of CPT1b, the predominant isoform in cardiac and skeletal muscle. Mitochondrial extracts from Expi293F cells transfected with CPT1a or CPT1b expression plasmids served as sources of catalytically active enzymes. The assay was validated using three previously confirmed CPT1b inhibitors: (R)-(+)-etomoxir, perhexiline, and malonyl-CoA. We then generated side-by-side inhibitory profiles for both isoforms, identifying vincamine as a lead selective inhibitor of CPT1b. Furthermore, chlorpromazine, previously characterized only as a broad CPT1 inhibitor and subsequently shown to inhibit CPT1a, is demonstrated here to also inhibit CPT1b, expanding its known isoform profile. Together, these results establish a robust platform for comparative isoform profiling and demonstrate that selective modulation of CPT1b is achievable, with implications for targeted therapeutics in metabolic and oncological disease.

cell biology↗

bioRxiv2026

GAE-Δ: A Graph-Learning Framework for Gene Network Rewiring and Clinical Outcome Prediction from Multi-Omics Data

Cancer progression and outcomes are driven in part by changes to molecular networks that result from genetic and/or environmental perturbations. These network changes manifest across multiple interconnected network layers and include accumulation of somatic mutations, altered proteinprotein interactions and dysregulated gene-expression. Here we describe a graph autoencoder-based framework (Graph Autoencoder-Delta (GAE-{Delta})), for characterizing phenotype-specific gene role shifts across multi-omics data. Given samples stratified into two contrasting phenotypic groups and a prior gene interaction network, GAE-{Delta} constructs group-specific gene graphs for each omics modality and trains, for each modality, a single graph autoencoder jointly on both group graphs, so that the two group-conditional embeddings share a common latent space. Contrasting these embeddings defines a multi-omics embedding-shift representation for each gene that reflects how its network role reorganizes across phenotypic contexts. These gene-level shifts are subsequently used for unsupervised gene prioritization, multi-omics late fusion and sample-level classification. Applied to five TCGA cancer types with a survival endpoint, GAE-{Delta} achieves competitive or superior predictive performance compared with classical network-based methods and multi-omics matrix-factorisation methods (MOFA+, iNMF), with statistically significant AUC gains over MOFA+ in three of five cohorts and statistical ties on the remaining two. Beyond predictive performance, the consensus shift genes are significantly enriched for known cancer drivers in three of five cohorts (hypergeometric p < 0.01; 11-17x fold-enrichment), whereas matrix-factorisation baselines reach p < 0.05 in zero of five cohorts (best per-cancer p = 0.06), indicating that GAE-{Delta} captures biological signal that linear factor methods miss. In summary, the GAE-{Delta} approach provides for both improved outcome classification as well as for biological and mechanistic discovery through deep network-based integration of disease-associated multi-omics data.

bioinformatics↗

bioRxiv2026

Multiomic State-Transitions Reveal Post-Treatment Transcriptome Desynchronization in Acute Myeloid Leukemia

Temporal dynamics of the peripheral blood transcriptome are crucial for understanding leukemia evolution and response to therapy because they can reveal how gene expression programs drive abnormal cell states, disease heterogeneity, and treatment resistance. Using a mathematical model of state-transitions, we studied the temporal dynamics of peripheral blood messenger RNA (mRNA) and microRNA (miRNA) transcriptomes in a mouse model of acute myeloid leukemia (AML). In our state-transition model, mRNA and miRNA transcriptomes are represented as a particle undergoing Brownian motion in a two-dimensional multiomic potential landscape. Following chemotherapy, we observed a temporal desynchronization between mRNA and miRNA transcriptomic responses corresponding to an asymmetric shift in the landscape. Specifically, mRNA trajectories responded almost immediately post-treatment, whereas miRNA responses were delayed by approximately two weeks. Clustering analysis identified that the temporal delay is driven by a prominent cluster of miRNAs from the imprinted Dlk1-Dio3 region. Although previously implicated in acute promyelocytic leukemia, lymphomas, and metabolic dysregulation, this provides the first evidence linking the Dlk1-Dio3 locus to AML chemotherapy response and treatment-induced transcriptomic desynchronization. This framework offers an innovative dynamics-based strategy to identify biological drivers of therapeutic response and novel therapeutic targets across hematological malignancies. Key PointsO_LITreatment can induce desynchronization between messenger RNA and microRNA transcriptome dynamics in a murine model of AML C_LIO_LIMicroRNAs from the imprinted Dlk1-Dio3 locus are identified as key contributors to desynchronization and may serve as therapeutic targets C_LI

cancer biology↗