Biology subjects

Liu, G.

Publications and source records attributed to Liu, G..

At least 91 records · Page 5Linked to original sources

bioRxiv2025

ZMS-6659, a selective Polθ helicase inhibitor, demonstrates anti-tumor activity as monotherapy and synergy effect with PARP inhibitors and topoisomerase inhibitors

The repair of DNA double strand break (DSB) is crucial for genome stability. Homologous recombination (HR) is a high-fidelity, error-free DSB repair pathway, and its dysfunction leads to genome instability and can result in tumorigenesis. In HR deficient contexts, microhomology-mediated end joining (MMEJ), dependent on DNA polymerase theta (Pol{theta}), is up-regulated to serve as a backup pathway for DSB repair. Several studies have proved that Pol{theta} inhibition causes synthetic lethality with HR deficiency, and Pol{theta} emerges as a potential DNA damage repair (DDR) target for the treatment of HR deficient tumors both as single agent or in combination with PARP inhibitors (PARPi). In addition, evidences show that the blockage of Pol{theta} can enhance the sensitivity to DNA damage agents. Herein, we characterized ZMS-6659, a highly potent small-molecule Pol{theta} inhibitor targeting the helicase domain. ZMS-6659 induces synthetic lethality in HR deficient models, elicits synergistic effects with even low dose of PARPi and delays PARPi resistance without additional hematotoxicity. Besides, ZMS-6659 also enhance the sensitivity to topoisomerase inhibitors in HR proficient cell lines. These findings support the clinical exploration of ZMS-6659 in combination with PARPi in HR deficient tumors, and suggest a potential application expansion beyond HR deficiency.

cancer biology↗

bioRxiv2025

Detection of Aggressive Mesenchymal Glioblastoma by Mannose-Weighted CEST MRI

Glioblastoma (GBM) contain mesenchymal cancer stem cells that drive tumor aggressiveness and recurrence and exhibit aberrant glycosylation during proneural-to-mesenchymal transition. A comprehensive analysis of human GBM transcriptomic datasets revealed an upregulation of 13 genes involved in mannosylation. Histopathological staining of a tissue array representing 35 GBM cases revealed elevated mannose, correlating with increased expression of the mesenchymal marker CD44. Mannose-weighted chemical exchange saturation transfer magnetic resonance imaging (MANw CEST MRI) detected elevated mannose levels in aggressive mesenchymal GBM neurospheres in vitro and in vivo, but not in less aggressive non-mesenchymal phenotype. To establish causation, inhibiting the expression of the mannose binding lectins LMAN1/2 that regulate intracellular processing of mannosylated proteins decreased the glioma cell MANw CEST MRI signal. Our findings indicate that MANw CEST MRI can visualize high mannose levels in mesenchymal GBM cells, which may serve as a surrogate imaging biomarker for predicting and assessing tumor aggressiveness and recurrence.

cancer biology↗

bioRxiv2025

SCARF: Single Cell ATAC-seq and RNA-seq Foundation model

Recent advances in single-cell multi-omics have provided unprecedented insights into gene regulation by jointly profiling transcriptomic (scRNA-seq) and chromatin accessibility (scATAC-seq) landscapes. However, the inherent heterogeneity and high dimensionality of these multimodal data present significant challenges for effective integration and downstream analysis. Foundation models have demonstrated strong representation learning capabilities for scRNA-seq or scATAC-seq data. So far, however, no model has been specifically developed for the integrative analysis of these two modalities. Here, we introduce SCARF, a single cell ATAC-seq and RNA-seq foundation model. SCARF is pre-trained on X-Omics, the largest curated collection of single-cell multi-omics data to date, comprising over 2.7 million cells across multiple tissues and species. The model utilizes a Mamba architecture for efficiently capturing long-context relationships between genes and between accessible regions. Modality-specific and shared features are learned by the model through self-supervised learning and contrastive learning, respectively. SCARF achieves state-of-the-art performance on multiple downstream tasks, including cell representation, cell matching, and cross-omics translation. Furthermore, SCARF enables few-shot cell type annotation, demonstrating strong generalizability across previously unseen datasets. These results highlight the power of foundation models for advancing integrative analysis of single cell multi-omics data, with broad applications in important tasks including cellular characterization, gene or genomic perturbation analysis, and regulation network analysis.

bioinformatics↗

bioRxiv2025

The break in the cycle: Inositol pyrophosphate fluxomics disentangled via mathematical modelling

This study investigates the metabolic pathways of inositol pyrophosphates (IPPs) in the yeast cell line {Delta}SPX and the human tumor cell line HCT116. Utilizing pulse-labelling experiments with 18O water and ordinary differential equation (ODE) models, we explore the synthesis and turnover of the highly phosphorylated IPP, 1,5-InsP8. Our findings challenge the notion that 1,5-InsP8 can be synthesized through distinct routes, revealing a linear reaction sequence in both systems. Employing model reduction via the profile likelihood method, we achieved statistically concise identifiability analysis that led to significant biological insights. In yeast, we determined that 1,5-InsP8 production primarily occurs through the phosphorylation of 5-InsP7, with the pathway involving 1-InsP7 deemed unnecessary as its removal did not compromise model accuracy. In HCT116 cells, 1,5-InsP8 synthesis is mainly driven by 1-InsP7, with variations observed across different experimental conditions. These results underscore the utility of model reduction in enhancing our understanding of metabolic pathways, challenging traditional views of IPP metabolism, and providing a framework for future investigations into the regulation and implications of linear IPP pathways in eukaryotic cells.

molecular biology↗

bioRxiv2025

Experimental transmission of African swine fever virus by Ornithodoros lahorensis (Acari: Argasidae)

African swine fever (ASF) is an acute, highly contagious infectious disease that affects both domestic and wild pigs, caused by the African swine fever virus (ASFV). ASFV is the only known DNA virus transmitted by arthropod vectors, with acute infection in pigs leading to morbidity and mortality rates as high as 100%.The virus can persist in a transmission cycle among wild boars, soft ticks, and domestic pigs. To date, nine Ornithodoros spp. have been confirmed to be capable of transmitting ASFV worldwide. However, the potential soft tick species capable of transmitting ASFV in China remain unclear. In this study, we compared the ability of Argasidae spp. Ornithodoros lahorensis and Argas persicus, as well as Ixodidae spp., Haemaphysalis longicornis, Rhipicephalus sanguineus and Dermacentor silvarum, to transmit ASFV via animal transmission experiments. The results revealed that O. lahorensis soft ticks, but not A. persicus or Ixodes, could act as competent vectors through transstadial and transovarial transmission of ASFV. The virus titers horizontally transmitted ASFV to ticks were 105.83, 106.59 and 104.31, HAD50/ml respectively. These viruses were detected in the nymphs developed from larvae, adults developed from nymphs, and larvae hatched from eggs by adults. Thus, O. lahorensis ticks are likely an important natural vectors of ASFV, although both mammalian and tick hosts are likely required for the maintenance of ASFV in the sylvatic cycle.

microbiology↗

bioRxiv2025

Magnetically Labeled iPSC-Derived Extracellular Vesicles Enable MRI/MPI-Guided Regenerative Therapy for Myocardial Infarction

Stem cell-derived extracellular vesicles (EVs) offer a promising cell-free approach for cardiovascular regenerative medicine. In this study, we developed magnetically labeled induced pluripotent stem cell-derived EVs (magneto-iPSC-EVs) encapsulated with superparamagnetic iron oxide (SPIO) nanoparticles for image-guided regenerative treatment of myocardial infarction, in which EVs that can be detected by both magnetic resonance imaging (MRI) and magnetic particle imaging (MPI). iPSC-EVs were isolated, characterized per MISEV2023 guidelines, and loaded with SuperSPIO20 nanoparticles using optimized electroporation conditions (300 V, 2 x 10 ms pulses), achieving a high loading efficiency of 1.77 ng Fe/106 EVs. In vitro results show that magneto-iPSC-EVs can be sensitively detected by MPI and MRI, with a detectability of approximately 107 EVs. In a mouse myocardial ischemia-reperfusion model, intramyocardially injected magneto-iPSC-EVs (2 x 109) were imaged non-invasively by in vivo MPI for 7 days and ex vivo MRI, with the presence of magneto-iPSC-EVs confirmed by Prussian blue staining. Therapeutically, both native and magneto-iPSC-EVs significantly improved cardiac function, with a 37.3% increase in left ventricular ejection fraction and 61.0% reduction in scar size. This study highlights the potential of magneto-iPSC-EVs as a cell-free approach for cardiovascular regenerative medicine, offering both non-invasive imaging capabilities and therapeutic benefits for myocardial repair.

bioengineering↗

bioRxiv2025

Validation of bronchial airway gene expression associated with bronchiectasis in nasal epithelium

ObjectivesExamination of bronchial epithelium-derived gene expression signature of bronchiectasis (BE) in nasal epithelium. MethodsWe studied 220 participants from the Detection of Early Lung Cancer Among Military Personnel study with bulk RNA-seq of nasal epithelium brushings. Gene set enrichment analysis (GSEA) was used to examine whether genes previously identified as increased or decreased in the bronchial epithelium of individuals with radiologic BE are significantly enriched among the genes most significantly altered in the nasal epithelium. GSEA and cell-type specific gene expression signatures were used to examine changes in the cellular composition of nasal epithelium associated with radiologic BE. ResultsGenes previously found to have significantly altered expression in bronchial epithelium in radiologic BE are concordantly enriched among the genes most increased or decreased in nasal epithelium in radiologic BE in this validation set (p = 3x10-7, increased; p = 6x10-8, decreased). GSEA performed using cell-type signatures demonstrates that in nasal epithelium, BE is associated with increased multiciliated and deuterosomal cell-related gene expression and decreased expression of basal cell- related gene expression, consistent with our previous findings in bronchial epithelium. ConclusionOur work validates our bronchial signature of BE in an independent dataset and demonstrates that similar gene expression changes are associated with radiologic BE in the bronchial and nasal epithelium. These findings support considering nasal epithelium brushing as a less invasive tool for screening and monitoring BE. SummaryThis study validates a previously reported bronchial signature of bronchiectasis (BE) in nasal epithelium. Changes in gene expression and cell-type composition associated with radiologic BE are similar across both bronchial and nasal epithelium.

bioinformatics↗

bioRxiv2025

Epithelial miR-149-5p up-regulation is associated with immune evasion in progressive bronchial premalignant lesions

The molecular drivers bronchial premalignant lesion progression to invasive lung squamous cell carcinoma are not well defined. Prior work profiling longitudinally collected bronchial premalignant lesion biopsies by RNA sequencing defined a proliferative subtype, enriched with bronchial dysplasia. We found that a gene co-expression module associated with interferon gamma signaling and antigen processing/presentation was down-regulated in progressive/persistent versus regressive lesions within the proliferative subtype, suggesting a functional impact of these genes on immune evasion. RNA from these same premalignant lesions was profiled by microRNA (miRNA) sequencing and a miRNA-gene network analysis identified hsa-miR-149-5p as a potential regulator of this antigen presentation gene co-expression module associated with lesion progression. hsa-miR-149-5p was found to be predominantly expressed in the epithelium and up-regulated in progressive/persistent versus regressive proliferative lesions while targets of this miRNA, the transcriptional coactivator of MHC-I gene expression, NLRC5, and the genes it regulates were down-regulated. MicroRNA in situ hybridization of hsa-miR-149-5p in tissue from adjacent fixed biopsies showed that hsa-miR-149-5p was increased in areas of bronchial dysplasia in progressive/persistent versus regressive lesions. Imaging mass cytometry showed that NLRC5 protein expression was decreased in progressive/persistent versus regressive lesions within areas of hyperplasia, metaplasia, and dysplasia. Additionally, basal cells with high versus low levels of NLRC5 were found to be in close spatial proximity to CD8 T cells, suggesting that these cells exhibit increased functional MHC-I gene expression in lesions with low hsa-miR-149-5p expression. Collectively, our data suggests a functional role for hsa-miR-149-5p in bronchial premalignant lesions and may serve as a therapeutic target for PML immunomodulation. STATEMENT OF SIGNIFICANCEIntegrative analysis across bronchial premalignant lesions has identified and localized a potential regulator of immune evasion in progressive/persistent lesions that could be a novel therapeutic target.

cancer biology↗

bioRxiv2025

Mouse Cortical Cellular Diversification Through Lineage Progression of Radial Glia

Cortical radial glia (RGs) sequentially generate pyramidal neurons (PyNs) and glia. In this study, we investigated the cell intrinsic programs underlying cortical cellular diversification using time-series scRNA-seq and snATAC-seq on purified mouse cortical progenitors across embryonic and postnatal stages. Our data revealed that RGs transition from early to late over time, sequentially producing intermediate neuronal progenitors (INPs) and intermediate glial progenitors (IGPs). While INPs expand exclusively to generate PyNs, IGPs progress from young to old, sequentially producing cortical astrocytes, oligodendrocytes, and olfactory bulb interneurons. We constructed comprehensive molecular maps that reflect cell lineage progression. In particularly, we found that chromatin accessibility drives cortical cellular diversification by restricting broadly expressed transcription factors to specific stages and cell types. Developmental changes in chromatin accessibility confine Lhx2-induced neurogenesis to early-stage RGs, leading to the loss of neurogenic competence and the acquisition of gliogenic competence as corticogenesis progresses.

neuroscience↗

bioRxiv2025

Nipah virus matrix protein utilizes cortical actin to stabilize the virus assembly sites and promote budding

Several families of enveloped viruses assemble and bud from the host cell plasma membranes (PM), including paramyxoviruses. Nipah virus (NiV) is a deadly zoonotic paramyxovirus causing yearly outbreaks in Southeast Asia with >75% mortality. NiV encodes matrix proteins (M) that drive assembly and budding. NiV-M forms dimers and interacts with membrane lipids in the host cells PM for budding. Using single-molecule localization microscopy and single-particle tracking, we show that the host F-actin maintains the nanoscale organization of NiV assembly sites at the PM. This F-actin-dependent integrity of NiV assembly sites is observed at the membrane retention stage after the assembly process is complete, rather than during the recruitment of NiV-M molecules to these sites. NiV-M interacts with actin via its C-terminal domain. We also show that the actin-branching factor, Arp2/3 complex, promotes virus-like-particle production. Meanwhile, inhibiting Arp2/3 disfavors the PM retention of NiV assembly sites and NiV-M-mediated generation of membrane protrusions, but does not affect the assembly rate. This suggests that Arp2/3-nucleated actin polymerization and branching are critical for maintaining NiV-M assembly sites at the PM, allowing it to facilitate membrane protrusion generation for virus budding. Our findings support the following model: NiV-M interacts with F-actin through its C-terminus to remain at the host PM after assembly completion. This F-actin-dependent retention is promoted by the Arp2/3-driven actin branching and polymerization, which also drives the formation of NiV-M-induced membrane protrusions necessary for virus budding. SignificanceNipah virus (NiV) is a deadly paramyxovirus capable of animal-animal and animal-human transmissions. To produce a NiV particle, the NiV matrix protein (M) must create an assembly site by binding to the surface of an infected cell and pushing the cell membrane outward for virus budding. To do so, the NiV matrix protein must co-opt or overcome an actin network underneath the cell membrane. We provide single-molecule evidence that M interacts with the actin cytoskeleton for membrane retention but not recruitment of M to existing assembly sites. This process is promoted by the Arp2/3-driven actin branching and polymerization. Our findings suggest the role of actin remodeling in NiV budding and identify a druggable site at the C-terminus of NiV-M.

microbiology↗

bioRxiv2025

Wearable Photonic Device for Multiple Biomarker Sampling and Detection Without Blood Draws

Blood draws or phlebotomy using needles have been practiced in clinics for several centuries and it often causes pain, discomfort, and inconvenience. Herein, we create a wearable photonic device by integrating a microlens array (MLA) and an optic microneedle array (OMNA) functionalized with immunobinding capacity for safe and needle-free biomarker sampling and detection. The MLA-integrated OMNA amplified and transmitted LED light at 595 nm into the skin through the OMNA, bypassing the light-absorbing melanin in the epidermal layer, and evenly distributing it in the capillary-enriched dermis. The 595 nm light was preferably absorbed by hemoglobin (Hb) and oxygen-Hb within the capillaries, triggering thermal dilation of capillaries without damaging them or causing petechiae. The light illumination resulted in remarkable increases in the concentrations of various blood biomarkers in the skin owing to capillary dilation and biomarker extravasation. These extravasated biomarkers bound specifically to the OMNA, which was covalently decorated with capture antibodies, each microneedle immobilized with one specific antibody. The multi-functional OMNA was extensively modified to amplify the immunobinding signals and achieve sensitivity superior to that of enzyme-linked immunosorbent assay (ELISA) kits. As a proof of concept, we validated the functionality of the prototype for minimally invasive sampling and precise multiplexed blood biomarker detection in two mouse models to quantify acute inflammation and specific antibody production. This cost-efficient device provides a promising platform for blood-free, multiplexed detection of blood biomarkers.

bioengineering↗

bioRxiv2025

Follicular mural granulosa cells stockpile glycogen to fuel corpus luteum pre-vascularization

The corpus luteum (CL) arises from the luteinization of follicular granulosa cells (GCs) and theca cells, marked by rapid progesterone elevation and angiogenesis. Intriguingly, angiogenesis lags behind progesterone elevation, creating an avascular phase during which luteal cells must fuel intensive steroidogenesis without perfusion. How the avascular CL meets this energetic demand remains a mystery. Here, we reveal a novel cellular adaptive mechanism-GC energy storage (GCES)-that resolves this enigma. We demonstrate that upon luteinization initiation, GCs enter a metabolically quiescent state yet enhance glucose uptake via SLC2A1, converting the glucose into glycogen through the hCG (LH)-MAPK-RUNX1-Insulin signaling axis. Catabolism of this glycogen reserve supplies the energy required for the avascular CL, ensuring normal luteogenesis. GCES is evolutionarily conserved across species. Genetic or pharmacological disruption of GCES or glycogenolysis induces luteal insufficiency, whereas timely glucose administration enhances GCES, improving luteal function and optimize reproductive outcome in both mouse and ovine models. In human study, orally intake of glucose post-hCG significantly augments GCES and enhances progesterone production in women. These results advance luteal physiology by uncovering a universal reproductive principle with direct clinical implications.

developmental biology↗

bioRxiv2024

Lotus japonicus VIH2 is an inositol pyrophosphate synthase that regulates arbuscular mycorrhiza

Plant yield is often maximized by the extensive use of mineral fertilizers, which, however, has severe environmental consequences. Phosphate is particularly problematic, as it represents a globally limited resource, and its runoff and soil erosion threaten open water bodies. Many crops engage in arbuscular mycorrhizal (AM) symbiosis with nutrient-acquiring fungi, aiding in the uptake of phosphate and other mineral nutrients. However, AM colonization is strongly reduced under high soil phosphate levels. A mechanistic understanding of phosphate sensing, phosphate starvation responses, and their connection to AM remains enigmatic. Here, we show that in Lotus japonicus, low-abundant, energy-rich inositol pyrophosphates act as master regulators of AM, orchestrating the crosstalk between phosphate starvation responses and plant root endosymbiosis. These findings hold promise for breeding nutrient-efficient crops.

plant biology↗

bioRxiv2024

Instantaneous Frequency: A New Functional Biomarker for Dynamic Brain Causal Networks

This study introduces instantaneous frequency (IF) analysis as a novel method for characterizing dynamic brain causal networks from fMRI blood-oxygen-level-dependent (BOLD) signals. Effective connectivity, estimated using dynamic causal modeling (DCM), is analyzed to derive IF sequences, with the average IF across brain regions serving as a potential biomarker for global network oscillatory behavior. Analysis of data from the Alzheimers Disease Neuroimaging Initiative (ADNI), Open Access Series of Imaging Studies (OASIS), and Human Connectome Project (HCP) demonstrates the methods efficacy in distinguishing between clinical and demographic groups, such as cognitive decline stages, sex differences, and sleep quality levels. Statistical analyses reveal significant group differences in IF metrics, highlighting its potential as a sensitive indicator for early diagnosis and monitoring of neurodegenerative and cognitive conditions. Graphical Abstract O_FIG O_LINKSMALLFIG WIDTH=200 HEIGHT=100 SRC="FIGDIR/small/628965v1_ufig1.gif" ALT="Figure 1"> View larger version (35K): org.highwire.dtl.DTLVardef@10da58org.highwire.dtl.DTLVardef@6eee6borg.highwire.dtl.DTLVardef@e6e756org.highwire.dtl.DTLVardef@15dff2a_HPS_FORMAT_FIGEXP M_FIG C_FIG HighlightsO_LIThe study introduces instantaneous frequency (IF) as a novel biomarker derived from dynamic brain effective connectivity, capturing temporal fluctuations in brain networks. C_LIO_LIThe proposed IF biomarker effectively differentiates between various clinical stages, such as Mild Cognitive Impairment (MCI) and Alzheimers Disease (AD), and demographic factors, including sex and sleep quality. C_LIO_LIThe robustness and clinical relevance of the IF biomarker are validated using three independent datasets: ADNI, OASIS, and HCP, demonstrating its potential in cognitive and neurological research. C_LI

neuroscience↗

bioRxiv2024

Interaction of osteoprotegerin with fibulin-1 is essential for extracellular matrix deposition in fibrotic lung tissue

Idiopathic pulmonary fibrosis (IPF) is a progressive, fatal lung disease with an unclear cause and no cure. Osteoprotegerin (OPG), a well-known regulator of bone extracellular matrix (ECM), is increased in pulmonary fibrosis and this is associated with lower lung function and IPF progression. Fibulin-1, a key ECM protein, is increased in lung fibroblasts and lung tissues from patients with IPF and high levels are predictive of a poor prognosis for these patients. We found a positive correlation between levels of OPG and fibulin-1 in serum from patients with IPF, but it is unknown if these proteins interact. This study aimed at investigating interactions between OPG and fibulin-1 in lung fibrosis, hypothesizing that OPG may form a complex with fibulin-1 thereby possibly influencing lung fibrosis. We investigated intracellular interactions between OPG and fibulin-1 in human lung fibroblasts, showing neither OPG nor fibulin-1 expression were dependent on each other. However, OPG protein deposition was significantly decreased in lung tissues of fibulin-1 knockout mice (fibulin-1-/-)compared to controls suggesting extracellular interactions. Subsequently, we detected colocalization of both proteins in both IPF and control lung tissues using immunofluorescence, with both of them also colocalizing with latent TGF{beta} binding protein 1 (LTBP1). Proximity ligation analyses confirmed close proximity of OPG to fibulin-1 and fibulin-1 to LTBP1, particularly in the interstitial regions of both control and IPF lung tissue, but not OPG and LTBP1. In conclusion, we found that OPG, fibulin-1 and LTBP1 seem to form a complex, with fibulin-1 playing a bridging role between OPG and LTBP1, in the extracellular environment. This complex may play a regulatory role in deposition of extracellular matrix in lung tissue.

pathology↗

bioRxiv2024

Multi-omics analysis of the mechanism of alfalfa and wheat-induced rumen flatulence in Xizang sheep

This study induced a rumen flatulence model in Xizang sheep using alfalfa (HRF) and wheat grass (MRF) to explore related mechanisms. The pH of HRF and MRF groups was lower than the natural grass group. SCFA concentrations varied in different groups. Microbial analysis showed HRF and MRF groups had lower alpha and beta diversity and different compositions. Transcriptome analysis found different numbers of differentially expressed genes in each group. Spearmans correlation analysis indicated Butyrivibrio was related to GLRX and DUOX2, and Quinella to several genes.

microbiology↗

bioRxiv2024

scCompass: An integrated cross-species scRNA-seq database for AI-ready

Emerging single-cell sequencing technology has generated large amounts of data, allowing analysis of cellular dynamics and gene regulation at the single-cell resolution. Advances in artificial intelligence enhance life sciences research by delivering critical insights and optimizing data analysis processes. However, inconsistent data processing quality and standards remain to be a major challenge. Here we propose scCompass, which provides a data quality solution to build a large-scale, cross-species and model-friendly single-cell data collection. By applying standardized data pre-processing, scCompass integrates and curates transcriptomic data from 13 species and nearly 105 million single cells. Using this extensive dataset, we are able to archieve stable expression genes (SEGs) and organ-specific expression genes (OSGs) in human and mouse. We provide different scalable datasets that can be easily adapted for AI model training and the pretrained checkpoints with state-of-the-art (SOTA) single-cell foundataion models. In summary, the AI-readiness of scCompass, which combined with user-friendly data sharing, visualization and online analysis, greatly simplifies data access and exploitation for researchers in single cell biology(http://www.bdbe.cn/kun).

bioinformatics↗

bioRxiv2024

Directing cellular transitions on gene graph-enhanced cell state manifold

A select few genes act as pivotal drivers in the process of cell state transitions. However, finding key genes involved in different transitions is challenging. To address this problem, we present CellNavi, a deep learning-based framework designed to predict genes that drive cell state transitions. CellNavi builds a driver gene predictor upon a cell state manifold, which captures the intrinsic features of cells by learning from large-scale, high-dimensional transcriptomics data and integrating gene graphs with directional connections. Our analysis shows that CellNavi can accurately predict driver genes for transitions induced by genetic, chemical, and cytokine perturbations across diverse cell types, conditions, and studies. By leveraging a biologically meaningful cell state manifold, it is proficient in tasks involving critical transitions such as cellular differentiation, disease progression, and drug response. CellNavi represents a substantial advancement in driver gene prediction and cell state manipulation, opening new avenues in disease biology and therapeutic discovery.

cell biology↗