Biology subjects

Li, Q.

Publications and source records attributed to Li, Q..

At least 73 records · Page 4Linked to original sources

bioRxiv2026

Tumor Microenvironment Modulates Lineage Plasticity in Lung Squamous Cell Carcinoma

Lung squamous cell carcinoma (LUSC) is the second most common type of lung cancer, yet therapeutic options remain limited. A deeper understanding of its biology and molecular pathogenesis is essential for developing new treatment strategies. Here, we investigated the mechanisms of phenotypic plasticity in LUSC by comparing organoid-derived orthotopic lung models (ODOLs) and subcutaneous xenograft models (ODXs). ODXs showed greater tumor growth, squamous differentiation, and extracellular matrix (ECM) organization compared to ODOLs. Transcriptomic analyses revealed upregulation of multiple HIF1 and SOX2 target genes together with enhanced hypoxia signaling in ODXs. CRISPR/Cas9-mediated HIF1-knockout ODXs showed reduced SOX2 expression, tumor growth, and ECM organization, whereas SOX2-knockout ODXs reduced tumor growth without affecting HIF1 and ECM organization. These results indicate that HIF1 regulates squamous lineage maintenance through SOX2 and ECM remodeling. Spatial transcriptomics revealed enrichment of basal cell-like and squamous-differentiated tumor states in ODXs, whereas ODOLs displayed less differentiated phenotypes. These findings identify the tumor microenvironment as a critical determinant of lineage plasticity in LUSC and provide mechanistic insight into how hypoxia shapes tumor differentiation.

Cancer Biology↗

bioRxiv2026

Comparison of Four Density Gradient Centrifugation Protocols for Purification of Toxoplasma gondii PRU Brain Cysts

Toxoplasma gondii is an obligate intracellular zoonotic protozoan that establishes persistent brain cysts in infected hosts, causing chronic infection and neuropathological damage. Efficient isolation and purification of brain cysts are essential for studying its biological characteristics and developing effective control strategies. The present study aimed to establish a reliable method for purifying brain cysts of the T. gondii PRU strain with high yield, viability, and practicality. Brain cysts harvested from experimentally infected ICR mice were purified using four different density gradient centrifugation methods: lymphocyte separation medium (LSM), Percoll, cesium chloride (CsCl), and sucrose. Purification efficiency was systematically evaluated, and the infection model was validated via brain histopathology. Cyst counts and purification yields were quantified, while cyst and bradyzoite viability were assessed using FDA/PI staining, trypan blue exclusion, and in vivo infectivity assays. Infected mice displayed the most severe clinical signs at 15 days post-infection (dpi), accompanied by significantly reduced body weight compared with uninfected controls (P < 0.05) and prominent perivascular inflammatory infiltration in the brain. Among the four purification methods, Percoll, CsCl, and sucrose gradients yielded significantly higher cyst numbers than LSM (P < 0.001), with no significant differences observed among the three gradient-based methods. Cysts purified by Percoll and LSM retained high viability and remained fully infectious in mice. Sucrose-purified cysts exhibited decreased viability, but their bradyzoites remained infective. In contrast, cysts and bradyzoites purified by CsCl were completely non-viable. These results clarify the advantages and limitations of each protocol and provide an optimized technical reference for T. gondii brain cyst research, supporting the development of toxoplasmosis prevention and control measures.

microbiology↗

bioRxiv2026

NF1 deficiency induces metabolic reprogramming and epithelial-mesenchymal transition in glioblastoma

BackgroundMetabolic reprogramming is a common occurrence in tumor cells, where enhanced glycolysis promotes cell growth, invasion and migration. NF1 is tumor suppressor gene that downregulates the encoded neurofibromin protein. However, the effects of NF1 on energy metabolism and epithelial-mesenchymal transition (EMT) in glioblastoma multiforme (GBM), as well as the underlying molecular mechanisms, remain unclear. MethodsCRISPR/Cas9 gene editing technology was employed to construct GBM cell lines with NF1 gene mutations. Metabolomics was utilized to examine the impact of NF1 on metabolic remodeling in GBM. The Seahorse XF24 extracellular flux analyzer was used to detect the effect of NF1 knockdown on glycolysis and mitochondrial oxidative phosphorylation in GBM cells. Wound healing assay and Transwell chamber assay were utilized to detect the effect of NF1 on GBM cell invasion. Orthotopic tumor model in nude mice was established to explore the role of NF1 in vivo. In addition, Co-IP, western blotting, and immunofluorescence were used to explore the changes of key enzymes in glycolysis and mitochondrial oxidative phosphorylation and the relationship between NF1 and MFN1. ResultsThe expression of NF1 is decreased in glioma tissues and is significantly correlated with patient prognosis. NF1 knockdown may promote the invasion, migration, and EMT of GBM cells. At the same time, the activation of the AKT/mTOR signaling pathway promotes aerobic glycolysis in GBM cells, promotes mitochondrial division through targeted regulation of MFN1, and inhibits mitochondrial oxidative phosphorylation. NF1 deficiency promotes EMT in GBM cells by enhancing aerobic glycolysis and mitochondrial division. ConclusionNF1 deficiency promotes GBM glycolysis by activating the AKT/mTOR signaling pathway and inhibits the mitochondrial oxidative phosphorylation by regulating MFN1; NF1 deletion promotes GBM EMT by remodeling the pattern of energy metabolism.

Cancer Biology↗

bioRxiv2026

Multitrophic interaction networks mediate biodiversity effects on ecosystem multifunctionality

Biodiversity loss threatens the multifunctionality of ecosystems on which human well-being ultimately depends. Changes in multitrophic species interactions may be key to explaining the ecological consequences of biodiversity loss, but research explicitly linking species interactions and ecosystem multifunctionality remains rare. To assess interaction-mediated biodiversity effects and underlying mechanisms, characterizing the structure of species interaction networks is invaluable. Using comprehensive species interaction and ecosystem functioning data from a large-scale tree biodiversity experiment, we find consistent effects of the structure of species interaction networks on ecosystem multifunctionality across multiple types of antagonistic and mutualistic interactions. While positive effects of network size align with expected positive effects of multitrophic species diversity, positive effects of niche overlap among interacting species and negative effects of highly connected species (i.e. high linkage density) reveal additional, interaction-mediated drivers of multifunctionality. Specifically, the effects of niche overlap suggest benefits of functionally similar species, and the effects of linkage density underscore the importance of specialized interactions in promoting ecosystem multifunctionality. These findings emphasize that to effectively safeguard ecosystem service provisioning, ecosystem management and biodiversity conservation not only need to account for biodiversity changes at multiple trophic levels, but also explicitly for how species interact among each other.

ecology↗

bioRxiv2026

Spatiotemporal Patterns and Structural Substrates of Individual Functional Variability in Youth

Youth is a period of emerging individuality and extensive neural remodeling, yet how functional brain individuality is organized across development remains unclear. Prior work has often conflated variability in functional topography with variability in functional connectivity, highlighting the need to examine them separately to better understand how functional individuality relates to brain structure and cognition. Here, using individualized functional parcellation in two large multimodal developmental cohorts (total N = 1211), we separately quantified variability in individualized functional parcellation (vIFP) and in functional connectivity (vFC). Both forms of variability were organized along the sensorimotor-association axis and were greatest in the association cortex. During development, vIFP increased significantly with age at both regional and whole-brain level, whereas vFC showed regionally specific age effects but no significant whole-brain increase. Both trajectories exhibited a common mid-adolescent inflection (14-16 years), marking a window of accelerated functional individualization. Despite shared spatial and temporal organization, vIFP and vFC showed dissociable associations with brain structure and cognition. vIFP was more strongly coupled to structural variability, whereas vFC was more strongly associated with cognitive variability. These findings reveal convergent and divergent developmental principles of topographic and connectional functional variability, highlighting their complementary roles in structural constraints and cognitive specialization.

neuroscience↗

bioRxiv2026

Structural Basis of Nidovirus Replication Organelle Evolution Revealed by the Arterivirus DMV Pore Complex

Positive-strand RNA viruses of the order Nidovirales - which include coronaviruses and arteriviruses - remodel host membranes into double-membrane vesicles (DMVs) as replication organelle to shield viral RNA synthesis from immune sensors. In coronaviruses, newly synthesized RNA is exported to the cytoplasm through a massive DMV pore complex formed by viral non-structural proteins (nsps). However, it remains unknown whether this elaborate architecture is unique to large-genome coronaviruses or a universal hallmark of the order. Here, we integrate in situ cryo-electron tomography and single-particle cryo-electron microscopy to resolve the atomic structure of DMV pore complex within its native membrane environment, from Equine Arteritis Virus (EAV), a prototype arterivirus with small genome. Despite lacking obvious sequence homology, the minimal EAV pore shares conserved architectural principles with its elaborate coronavirus counterpart. EAV pore complex is formed by a 12:12 stoichiometry of nsp2 and nsp3 protomers organized into four stacked concentric rings on double-membrane, yet generating pronounced structure symmetry mismatch. Functionally, the complex displays a distinct pore profile while preserving a positively charged central channel, essential for viral replication and transport. These findings demonstrate that, despite diversity in genome size and virion morphology, nidovirus replication organelles exhibit striking evolutionary conservation at the atomic structural level. Collectively, we propose that the order Nidovirales can be unified at the ultrastructural level by this conserved signature pore complex on the DMV-based replication organelle.

microbiology↗

bioRxiv2026

HoloCell: A Generative Foundation Model for Holistic Cellular Modeling

Single-cell multi-omics technologies have recently advanced to enable the profiling of epigenomic, transcriptomic, and proteomic layers within individual cells, offering new opportunities to characterize cellular states as integrated biological systems. However, developing a unified framework that can seamlessly integrate diverse omics modalities and remain robust to heterogeneous modality missingness remains challenging. Existing methods are often designed for specific modalities or modality pairs, relying on dataset-specific training or paired measurements. Here we present HoloCell, to our knowledge the first generative foundation model for joint representation learning and generative modeling across all three major single-cell omics modalities, i.e., epigenomics, transcriptomics, and proteomics. HoloCell contains over 860 million parameters and is pretrained on the Human-Multi-Omics-Corpus, which comprises approximately 468 million single-cell profiles across these three omics layers, corresponding to over 425 billion tokens. HoloCell introduces a a simple yet biologically motivated hierarchical tokenization strategy that encodes cis-regulatory elements, genes, and proteins as structured tokens within a shared modeling framework. We evaluated HoloCell across single-omics representation learning, paired multi-omics integration, unpaired multi-omics alignment, and cross-modal generation via iterative diffusion and remasking, demonstrating its superior performance and flexibility across diverse omics tasks. From a representation perspective, HoloCell provides a unified digital mapping of cellular states across multiple omics layers, capturing cell heterogeneity as an integrated system. From a generation perspective, its iterative diffusion and remasking frame-work permits flexible generation orders beyond fixed left-to-right causality, enabling in silico simulation of multi-omics information flow. Together, these capabilities position HoloCell as a versatile foundation model toward the emerging concept of a virtual cell, offering both systematic characterization and generative simulation of cellular systems within a unified framework.

bioinformatics↗

bioRxiv2026

Enhanced Target Binding by Leritrelvir Restores Dimerization of Mpro Mutants and Mitigates Drug Resistance

The SARS-CoV-2 main protease (Mpro) has been a major target of antiviral drug development, leading to the development of inhibitors such as nirmatrelvir, the antiviral component of the COVID-19 drug Paxlovid. However, resistance-associated mutations that reduce the efficacy of current Mpro inhibitors, particularly nirmatrelvir, have emerged. Here, we evaluated the inhibitory activity of leritrelvir (RAY1216), an Mpro inhibitor approved in China for COVID-19 monotherapy, against a panel of Mpro variants carrying mutations at 12 resistance-associated residues distributed across four catalytic subsites. Using integrated biochemical, biophysical, structural, and cellular analyses, we demonstrate that leritrelvir retains stronger inhibitory activity against most tested resistant mutants compared with nirmatrelvir. Most of the tested mutations promote Mpro dimer dissociation, with E166V showing a particularly pronounced effect and markedly compromising nirmatrelvir binding. In contrast, thermal shift and size-exclusion chromatography assays demonstrate that leritrelvir binding restores dimerization of these Mpro mutants. Sixteen high-resolution crystal structures reveal that leritrelvir binding re-instates key dimer-interface interactions disrupted by resistance mutations. Mini-replicon assays further confirm leritrelvir to possess enhanced cellular antiviral efficacy compared with nirmatrelvir. Our findings indicate that tighter leritrelvir binding enables more effective inhibition of dissociation-prone Mpro mutants than nirmatrelvir, supporting its use as a more resilient antiviral agent for SARS-CoV-2 treatment.

molecular biology↗

bioRxiv2026

X-Gutor: a high-fidelity in vitro platform for long-term gut microbiome recovery

In vitro gut microbiome-culturing models are essential for studying host-microbe interactions, yet achieving high microbial recovery and long-term stability remains a long-standing challenge. Here we devise X-Gutor, an ecologically designed in vitro platform that recapitulates the monogastric digestive system with precise control of pH, redox potential, and a mucin-based biofilm matrix. Using an integrated assessment framework, microbial source tracking, a curated gut metabolism database (GutDB), and short-chain fatty acid profiling, we benchmarked gut community recovery. Building on this platform, HanGutor and PigGutor were developed to recover 96.27{+/-}4.92% and 90.48{+/-}5.95% of human and swine gut microbiota, respectively. Ecological fine-tuning revealed that redox potential drives the Prevotella-Bacteroides trade-off and that spatial biofilm structure suppresses cheater (Succinivibrio) growth. The X-Gutor provides a scalable, modular platform for gut microbiome engineering, dietary intervention, and microbiota-targeted therapy development.

microbiology↗

bioRxiv2026

ACC1-Dependent De Novo Lipogenesis Sustains Hematopoietic Stem Cell Quiescence and Self-Renewal

Certain aspects of lipid metabolism are known to regulate hematopoietic stem cell (HSC) function, including fatty acid oxidation and lipid uptake, but there is a limited understanding of the contribution of de novo fatty acid synthesis to HSC homeostasis. Here, we show that endogenous fatty acid synthesis is essential for HSC function. Conditional deletion of Acaca, the gene that encodes the rate-limiting enzyme for de novo fatty acid synthesis, acetyl-CoA carboxylase 1 (ACC1), in hematopoietic cells profoundly reduces HSC function, marked by a reduced ability to reconstitute irradiated mice after competitive transplantation. ACC1 deficiency reduced quiescence, increased uptake of extracellular lipids, and increased reactive oxygen species in HSCs. The loss of HSC function is partly caused by increased fatty acid oxidation as deletion of CPT1a, which is required for long-chain fatty acid oxidation, partially rescued HSC function. A balance between fatty acid synthesis and fatty acid oxidation is thus critical for the maintenance of HSC function.

cell biology↗

bioRxiv2026

Identification of highly immunogenic endogenous dsRNAs from cellular MDA5 filaments

ADAR1 converts adenosine to inosine in endogenous double-stranded RNAs (dsRNAs) to prevent excessive MDA5-driven interferon-stimulated gene expression. The source of endogenous immunogenic dsRNAs remains enigmatic because only a small fraction of ADAR1 substrates activate MDA5, and cellular MDA5 filaments have not been isolated. Here, we couple affinity purification of cellular MDA5 filaments with RNA sequencing to define immunogenic endogenous dsRNAs. Greater than 84% of dsRNAs suppressed by combined DDX3X RNA helicase and ADAR1 base-editing activities were present in MDA5 filaments, compared to less than 1% of dsRNA substrates acted on by ADAR1 alone. Dual substrate dsRNAs consisted of inverted repeats embedded in 3-UTRs with high base-pair complementarity and longer intervening sequences between repeats, with a minor contribution coming from intermolecular dsRNAs formed by sense and antisense transcripts. Moreover, the majority of dual substrate immunogenic dsRNAs were hyperedited in DDX3X mutant cancers. This reveals the identity of endogenous immunogenic dsRNAs and quality control mechanisms underlying their suppression.

cell biology↗

bioRxiv2026

TetraFuse: A Synergistic Four-Dimensional Dynamic Fusion Framework for Efficient and Robust Medical Image Classification

Accurate and robust classification of medical pathology images is pivotal for computer-aided diagnosis. However, the deployment of deep learning models in high-throughput clinical screening faces a fundamental challenge: the trade-off between diagnostic accuracy and computational efficiency. Current lightweight architectures, while reducing parameter complexity through grouped convolutions, often lead to cross-channel information isolation and diminished representational capacity. In this paper, we propose TetraFuse, a novel framework that systematically integrates features from four complementary domains: space, channel, statistics, and frequency. TetraFuse introduces a novel Cross-Channel Dynamic Aggregation (CCDA) paradigm that reconstructs global channel topology with negligible computational overhead, resolving the inter-group isolation issue. To balance perceptual fidelity and efficiency, we design a stage-aware local enhancement mechanism: Local Variance-Guided Enhancer (LVGE) is employed to filter out shallow-stage background noise, while High-Frequency Boundary Injection (HFBI) reinforces deep-stage pathological contours, preventing spatial over-smoothing. Experimental results on the COVID-19, ISIC 2018, and Kvasir datasets confirm that TetraFuse outperforms state-of-the-art (SOTA) methods. Notably, TetraFuse-Tiny achieves a transformative 91.53% reduction in FLOPs compared to ResNet50; on the Kvasir dataset, it achieved an accuracy of 0.926 and an AUC of 0.994 with only 0.345G FLOPs. By combining high representational power with minimal computational demand, TetraFuse offers a scalable solution for large-scale medical image analysis, especially in resource-constrained clinical environments.

bioinformatics↗

bioRxiv2026

Extracellular salicylic acid activates immune signaling through cell-surface receptors

Salicylic acid (SA) is a central immune hormone that accumulates in both intracellular and extracellular compartments during pathogen infection. While intracellular SA signaling is well established, whether extracellular SA (eSA) directly activates immune responses remains unknown. Here we show that eSA functions as an extracellular signal potentially perceived by the plasma membrane-localized lectin receptor kinases LecRK-I.8 and LecRK-VI.2 in Arabidopsis. The extracellular domains of both receptors bind SA with micromolar affinity, SA rapidly induces LecRK-I.8 phosphorylation, and the kinase activities of LecRK-I.8 and LecRK-VI.2 are required for downstream signaling. Moreover, mutagenesis of a computationally predicted binding pocket in LecRK-VI.2 abolishes SA binding and immune function, providing evidence for receptor-mediated perception. Genetic analyses further demonstrate that LecRK receptors are required for SA-induced resistance, transcriptional reprogramming, and phosphoproteomic responses. Together, these findings expand current models of SA signaling by revealing a potential receptor-mediated cell-surface perception mechanism for a classical immune hormone.

plant biology↗

bioRxiv2026

inGSEA: An Improved Method for Gene Set Enrichment Analysis Using a Weighted Integral Statistic

Gene Set Enrichment Analysis (GSEA) is one of the most popular methods for transcriptomic analysis, yet its statistical power is limited when the biological pathways exhibit heterogeneous or non-concordant expression patterns. We propose an improved GSEA method, integral-based GSEA (inGSEA). inGSEA introduces a novel enrichment score based on the Anderson-Darling weighted integral statistic. The new enrichment score enhances detection power for complex signals, particularly sparse and bidirectional ones, while the Cauchy combination of integral and classic maximum statistics provides robustness across diverse expression patterns. Extensive numerical studies demonstrate that inGSEA achieves superior power and well-calibrated false discoveries. Application to real-world datasets reveals biologically relevant pathways missed by the standard GSEA. inGSEA reduces the computational burden of permutation testing by employing a generalized gamma distribution to approximate the null distribution. inGSEA is accessible as a user-friendly web-based software tool (https://amss-stat.github.io/inGSEA).

bioinformatics↗

bioRxiv2026

A unified developmental framework of the human placenta in its uterine environment in vivo and in vitro

Despite advances in single-cell profiling of the human placenta, the genetic programs governing its physiological remodeling throughout gestation remain incompletely understood; this limits the interpretation of trophoblast organoid models. Here, we reconstruct the human placenta in its uterine environment across gestation by integrating public single-cell data into a unified developmental framework developed through a specialized computational strategy. We resolve 100 cell subtypes, expanding the known cellular repertoire and uncovering extensive gestational dynamics. In the placental mesenchymal core, we define a stromal-vascular niche comprising previously unresolved fibroblast heterogeneity and vascular hierarchies (capillary, arterial, and venous). This niche undergoes reprogramming from early angiogenesis to vascular maturation at term and engages signaling programs that support villous homeostasis. Within the trophoblast lineage, we uncover differential progenitor dynamics: bipotent cytotrophoblast (CTB) progenitors persist throughout gestation, whereas extravillous trophoblast (EVT)-biased progenitors are almost absent at term, coinciding with differentiation into specialized states. Benchmarking trophoblast organoids against this reference shows distinct regional identities and developmental biases. Tissue-derived models recapitulate villous CTB whilst trophoblast stem cell-derived organoids resemble smooth chorion CTB; all models capture early gestational syncytiotrophoblast and progressive EVT differentiation. Together, this work provides a resource for understanding placental remodeling across gestation and guiding the use of in vitro models.

cell biology↗

bioRxiv2026

Structural Dynamics of RNA Polymerase II During Nucleotide Addition Cycle

RNA polymerase II (RNAPII) drives gene expression through iterative nucleotide addition cycles (NACs) comprising translocation, substrate binding, and catalysis. The lack of pre-catalysis and post-catalysis intermediates has precluded a complete mechanistic understanding of the NAC. Here we present 43 cryo-EM structures capturing distinct stages of the S. cerevisiae RNAPII elongation complex (EC) NAC, including previously intractable transition intermediates. We establish a continuous spectrum of RNAPII EC structural dynamics during the NAC, which can be divided into two coordinated phases: a substrate-induced EC tightening phase and a post-catalysis EC relaxation phase. For the substrate-induced EC tightening phase, the substrate binding initiates allosteric conformational changes across the entire RNAPII EC, including TL folding, funnel closure, clamp closure, transcription bubble ordering, and precise alignment of the RNA 3'-end with substrate to form a catalysis-competent configuration. For the post-catalysis EC relaxation phase, we captured the long-sought, short-lived post-catalysis product state and identified a series of intermediates that reveal a reverse conformational transition that facilitates rapid translocation. Together, our findings define a comprehensive structural and dynamic framework for RNAPII NAC, yielding a "molecular movie" of RNAPII in action and revealing a fundamental principle by which the enzyme balances speed and fidelity through coordinated conformational dynamics.

molecular biology↗

bioRxiv2026

Genome Assembly of the Green Revolution Wheat Cultivar Pavon 76 Establishes a Reference for CIMMYT-Derived Wheat

Common wheat (Triticum aestivum L.) is a cornerstone of global food security. However, the standard reference genome, IWGSC RefSeq v2.1, is derived from wheat cultivar Chinese Spring, a historically important lineage that is not representative of modern cultivated wheat. In contrast, Pavon 76 is a CIMMYT Green Revolution wheat cultivar grown worldwide and is present in the pedigrees of numerous elite wheats, as well as in the genetic background of over 600 cytogenetic stocks used in studies ranging from recombination to male sterility. As no Pavon 76 reference has been reported, interpretation of these Pavon 76-derived stocks has relied on the genetically distant IWGSC RefSeq v2.1. To address this gap, we generated PacBio HiFi reads and Hi-C read pairs to assemble Pavon 76s genome. The final assembly comprises the expected 21 chromosomes, spans 15 Gb, and has an N50 of 710 Mb. Annotation identified 167,201 genes, of which 89.5% were functionally annotated. Comparative sequence and gene synteny analysis between Pavon 76 and IWGSC RefSeq v2.1 revealed significant structural divergence, establishing that IWGSC RefSeq v2.1 is not fully representative of modern wheats. This robust Pavon 76 assembly establishes the first chromosome-scale de novo assembly of a CIMMYT-derived wheat and provides a foundation for future genetic and genomic studies.

genomics↗

bioRxiv2026

Spatial Gene Set Enrichment Analysis with Applications to Spatially Resolved Transcriptomic Data

Spatially resolved transcriptomics enables the systematic characterization of spatial gene expression variation across tissue sections. Spatially variable genes within the same biological pathway often exhibit similar spatial expression patterns, reflecting shared biological functions and tissue organization. However, existing gene set enrichment analysis methods typically ignore this spatial dependence, which may reduce power to detect spatially organized pathways and limit the interpretability of pathway-level findings. To address this limitation, we propose spaGSE, a Bayesian hierarchical model for spatial pathway enrichment analysis that integrates genelevel summary statistics from spatial expression analysis with predefined gene set annotations. spaGSE models latent spatially variable gene signals through a Gaussian mixture framework and links spatial variation to gene set membership using logistic regression. To support robust and interpretable inference, we impose a spike-and-slab prior on the enrichment coefficient. Through simulation studies and analyses of four public SRT datasets, we show that spaGSE is scalable and achieves higher power while maintaining false positive rate control compared with existing approaches. In real-data applications, spaGSE identifies biologically relevant pathways with coordinated spatial organization across cancer and developmental tissues, demonstrating the value of incorporating spatial information into pathway-level inference for spatial transcriptomics.

bioinformatics↗