Biology subjects

Tan, K.

Publications and source records attributed to Tan, K..

At least 37 records · Page 2Linked to original sources

bioRxiv2025

E2f coordinates the cell cycle and cell fate of hematopoietic progenitors to drive stress myelopoiesis

During inflammation, cytokines such as Tnf, Il1{beta} and Il6 stimulate hematopoietic stem and progenitor cells (HSPCs) to proliferate and accelerate the production of inflammatory cells from the myeloid lineage through a process called stress myelopoiesis. Genetic inactivation of repressors of cell cycle activity in HSPCs is sufficient to recapitulate stress myelopoiesis, suggesting that the cell cycle activity and cell fate decision of proliferative HSPCs are coordinated by a cell intrinsic mechanism. However, the nature of this mechanism remains unknown. Here, we show that E2f simultaneously regulates proliferation of HSPCs, repression of alternative cell fates via the activation of Suz12/Ezh2-containing PRC2 complex and promotion of the myeloid fate by enhancing the signaling activity of CD131, the common {beta}chain receptor for {beta}cytokines (i.e. Il3 and Gm-Csf). Accordingly, dual inhibition of Ezh2 and {beta}cytokine signaling activity in a preclinical model of colitis represses stress myelopoiesis and restores colon homeostasis. Our results suggest that dual targeting of {beta}cytokine signaling and Ezh2 activity represents a novel therapeutic strategy to repress the production of pro-inflammatory myeloid cells in a wide spectrum of inflammatory diseases.

cell biology↗

bioRxiv2025

Scaling law links plant growth variation to grain yield in wheat stands

Growth rate, a fundamental biological trait influencing plant resource use, scales predictably with plant mass, following remarkably consistent allometric power laws shaped by biophysical constraints and natural selection. However, how these laws apply to crop plants shaped by artificial selection, and how they manifest in agronomic traits, remains undefined. Under controlled greenhouse conditions, we quantified the relationship between plant mass and growth rate in 195 European winter wheat cultivars. We uncovered genetic variation in allometry linked to plant size, where increased leaf allocation and faster development elevated allometric exponents. Phenotypic and genetic analyses revealed adaptive strategies, ranging from large, slow-growing genotypes that support reproductive initiation to small, fast-growing genotypes that enhance reproductive effort. A shared genetic basis--associated with Photoperiod response-1 (Ppd-1)--linked growth allometry in the greenhouse to genotype-by-environment interactions for grain yield in field trials. This variation in growth allometry, shaped by breeding in diverse environments, reflects strategies that enhance adaptation to weather conditions. Our findings demonstrate that growth allometry is biologically robust and agronomically important because it scales to wheat yield under diverse, realistic field conditions.

plant biology↗

bioRxiv2025

Engineering novel AAV capsids by global de-targeting and subsequent muscle-specific tropism in mice and NHPs

Recombinant adeno-associated viral (rAAV) vectors are a potent tool, but their clinical application is restricted by insufficient target tissue transduction and liver toxicity. We employed a novel two-step engineering strategy to create novel rAAV capsids with global tissue de-targeting, then produced strong tissue-specific expression by adding a peptide sequence. We created a novel capsid, AAV.Zero1, with globally de-targeted transduction by loop swapping domains from AAV9 into AAV2. Making an R585A substitution (AAV.Zero2) re-targeted tissues but deleting residues 585-587 (AAV.Zero3) abrogated transduction. Inserting a myogenic peptide into AAV.Zero3 produced a novel capsid (AAV.eM) with strong muscle-specific transgene expression while maintaining minimal off-target expression, including in liver, which was conserved in two mouse strains and non-human primates. AAV.eM showed similar expression as the leading myotropic vector MyoAAV.4A but had a more favorable safety profile. Importantly, AAV.eM was able to functionally rescue a mouse model of Duchenne Muscular Dystrophy following systemic delivery of a micro-dystrophin gene. Thus, AAV.eM is an improved myotropic rAAV capsid that de-targets other tissues, especially the liver, and proof-of-concept for a platform to create capsids with specific properties that translate across species by addition of peptides onto low transduction backbones.

bioengineering↗

bioRxiv2025

Application of spatial transcriptomics across organoids: a high-resolution spatial whole-transcriptome benchmarking dataset

Stem cell-derived organoids hold promise to model tissue-specific disease. To enable this, it is crucial to assess how transcriptional signatures, cellular organisation and composition of organoids compare to in vivo counterparts. However, technologies which elucidate regional molecular identity, like spatial transcriptomics, have been challenging to apply to organoids. This study presents the first systematic profiling of multiple stem cell derived organoid models (brain, heart muscle, heart valve, kidney, lung, cartilage, and haematopoietic) with Stereo-seq, a full transcriptome, spatial transcriptomics assay using on-chip in situ RNA capture at subcellular resolution. It describes optimisation of this assay to characterise organoids, use of multiple organoid samples on a single chip, assess differences in RNA capture efficiency compared to reference tissues and its limitations. This study introduces a bespoke analysis method that partitions samples into regions and further characterises them. These findings inform future works to characterise organoids using spatial transcriptomics, providing insights in optimising RNA capture of multiple organoids across a chip and novel methods for regional analysis.

bioinformatics↗

bioRxiv2025

N-Orbit: towards a universal model and metric for comparing tissue microenvironments

Advances in spatial omics technologies have greatly expanded our ability to explore tissue microenvironments (TMEs) across development and disease. Yet a theoretical framework for modeling and comparing tissue architecture in diverse biological contexts remains lacking. We introduce, N-Orbit, a mathematical model that captures both cell type composition and spatial relationships within tissue cellular neighborhoods, encoding this information in a vector format for computationally efficient distance metric calculations. While not a neighborhood detection method itself, N-Orbit enhances insights gleaned from the neighborhood generated by the plethora of recently developed methods. We benchmarked the N-Orbit-based neighborhood distance metric on spatial omics datasets containing ground truth neighborhoods and patient clinical outcomes. We demonstrated that the N-Orbit outperforms cell type enrichment (CTE)-based metrics in discriminating neighborhood types, predicting clinical variables, and identifying homologous structures across species. Additionally, N-Orbit enhances model interpretability by allowing neighborhoods to be traced back to their enriched N-Orbit structures. N-Orbit holds significant potential for deepening our understanding of how TMEs remodel during development, disease, and evolution.

bioinformatics↗

bioRxiv2025

Target engagement studies and kinetic live-cell degradation assays enable the systematic characterization of HDAC6 PROTACs at endogenous protein expression levels

Histone deacetylase 6 (HDAC6) is an important drug target for the treatment of cancer, inflammation, and neurodegenerative disorders. In recent years, the development of proteolysis-targeting chimeras (PROTACs) has emerged to achieve the chemical knockdown of HDAC6. Consequently, there is an urgent need to develop efficient methods for target engagement studies and to enable a thorough characterization of the degradation efficiency and kinetics of HDAC6 PROTACs. In this work, we present a simple NanoBRET assay to assess HDAC6 cellular target engagement using a HeLaHDAC6-HiBiT cell line that stably expresses the LgBiT protein. For this purpose, we successfully designed, synthesized, characterized, and utilized the cell permeable TAMRA-based fluorescent ligand 5. The key advantage of this NanoBRET assay using HeLaHDAC6-HiBiT cells is the endogenously tagged HDAC6, allowing us to study binding of inhibitors in a near-native environment. Furthermore, we succeeded in establishing a system for kinetic live cell monitoring of HDAC6 degradation. The analysis of the degradation kinetics of a set of HDAC6 PROTACs provided detailed insights into their degradation efficiency and will be helpful for the development of improved HDAC6 degraders in the future. O_FIG O_LINKSMALLFIG WIDTH=200 HEIGHT=98 SRC="FIGDIR/small/646177v2_ufig1.gif" ALT="Figure 1"> View larger version (27K): org.highwire.dtl.DTLVardef@147069forg.highwire.dtl.DTLVardef@3d64eorg.highwire.dtl.DTLVardef@1eac700org.highwire.dtl.DTLVardef@aedc3d_HPS_FORMAT_FIGEXP M_FIG C_FIG

pharmacology and toxicology↗

bioRxiv2024

An epigenetic basis for sustained inflammatory epithelial progenitor cell states in Crohn's disease

Defining consequential differences in intestinal epithelial stem cells in healthy humans versus those with inflammatory bowel disease (Crohns disease and ulcerative colitis) is essential for the development of much needed therapies to restore the epithelial barrier and maintain its fidelity. Employing single cell transcriptomic/epigenomic approaches and colonoid models from children and adults with Crohns disease led us to identify an inflammatory secretory progenitor (ISP) cell state present almost exclusively in patients with Crohns disease compared to control subjects. ISPs exhibit gene expression profiles consistent with normal secretory progenitor cells but concomitantly express a suite of distinguishing pro-inflammatory genes. Mechanistically, ISPs exhibit open chromatin and gene expression of ISP signature genes. While these ISP-specific genes are not expressed in intestinal stem cells, their chromatin is accessible in Crohns disease stem cells suggesting that ISP genes are epigenetically poised in stem cells and are transcriptionally activated in ISP cells in the presence of inflammatory stimuli. Consistently, Crohns disease colonoids exhibit sustained ISP gene expression that can be elicited further with pro-inflammatory cytokines or via co-culture with pro-inflammatory macrophages. In summary, we define differences in the epithelial stem and progenitor compartment of patients with Crohns disease suggesting aberrant stem cell differentiation and inflammatory gene expression arises during disease. HIGHLIGHTSO_LIWe identify an inflammatory secretory progenitor cell state in endoscopically non-inflamed tissue from pediatric and adult patients with Crohns disease. C_LIO_LICrohns disease epithelial stem and progenitor cells display increased chromatin accessibility preceding the emergence of inflammatory secretory progenitor cell states. C_LIO_LICrohns disease colonoids from non-inflamed regions contain inflammatory secretory progenitor cells in the absence of inflammatory stimuli C_LIO_LIInflammatory secretory progenitor cells expand in colonoid culture in response to cytokine stimulation or co-culture with pro-inflammatory macrophages C_LI

physiology↗

bioRxiv2024

A tumor-secreted protein utilizes glucagon release to cause host wasting

Tumor-host interaction plays a critical role in malignant tumor-induced organ wasting across multiple species. Despite known regulation of regional wasting of individual peripheral organs by tumors, whether and how tumors utilize critical host catabolic hormone(s) to simultaneously induce systemic host wasting, however, is largely unknown. Using the conserved yki3SA-tumor model in Drosophila, we discovered that tumors increase the production of adipokinetic hormone (Akh), a glucagon-like catabolic hormone, to cause systemic host wasting, including muscle dysfunction, lipid loss, hyperglycemia, and ovary atrophy. We next integrated RNAi screening and Gal4-LexA dual expression system to identify that yki3SA-gut tumors secrete Pvf1 to remotely activate its receptor Pvr in Akh-producing cells (APCs), ultimately promoting Akh production. The underlying molecular mechanisms involved the Pvf1-Pvr axis that triggers Mmp2-dependent ECM remodeling of APCs and enhances innervation from the excitatory cholinergic neurons. Interestingly, we also confirmed the similar mechanisms governing tumor-induced glucagon release and organ wasting in mammals. Blockade of either glucagon or PDGFR (homolog of Pvr) action efficiently ameliorated organ wasting in the presence of malignant tumors. Therefore, our results demonstrate that tumors remotely promote neural-associated Akh/glucagon production via Pvf1-Pvr axis to cause systemic host wasting.

physiology↗

bioRxiv2024

CelloType: A Unified Model for Segmentation and Classification of Tissue Images

Cell segmentation and classification are critical tasks in spatial omics data analysis. We introduce CelloType, an end-to-end model designed for cell segmentation and classification of biomedical microscopy images. Unlike the traditional two-stage approach of segmentation followed by classification, CelloType adopts a multi-task learning approach that connects the segmentation and classification tasks and simultaneously boost the performance of both tasks. CelloType leverages Transformer-based deep learning techniques for enhanced accuracy of object detection, segmentation, and classification. It outperforms existing segmentation methods using ground-truths from public databases. In terms of classification, CelloType outperforms a baseline model comprised of state-of-the-art methods for individual tasks. Using multiplexed tissue images, we further demonstrate the utility of CelloType for multi-scale segmentation and classification of both cellular and non-cellular elements in a tissue. The enhanced accuracy and multi-task-learning ability of CelloType facilitate automated annotation of rapidly growing spatial omics data.

bioinformatics↗

bioRxiv2024

Spatial and Single Cell Mapping of Castleman Disease Reveals Key Stromal Cell Types and Cytokine Pathways

Castleman disease (CD) is inflammatory lymphoproliferative disorder of unclear etiology. To determine the cellular and molecular basis of CD, we analyzed the spatial proteome of 4,485,009 single cells, transcriptome of 50,117 single nuclei, immune repertoire of 8187 single nuclei, and pathogenic mutations in Unicentric CD, idiopathic Multicentric CD, HHV8-associated MCD, and reactive lymph nodes. CD was characterized by increased non-lymphoid and stromal cells that formed unique microenvironments where they interacted with lymphoid cells. Interaction of activated follicular dendritic cell (FDC) cytoplasmic meshworks with mantle zone B cells was associated with B cell activation and differentiation. VEGF, IL-6, MAPK, and extracellular matrix pathways were elevated in stromal cells of CD. CXCL13+ FDCs, PDGFRA+ T-zone reticular cells (TRC), and ACTA2-positive perivascular reticular cells (PRC) were identified as the predominant source of increased VEGF expression and IL-6 signaling in CD. VEGF expression by FDCs was associated with peri-follicular neovascularization. FDC, TRC and PRC of CD activated JAK-STAT, TGF{beta}, and MAPK pathways via ligand-receptor interactions involving collagen, integrins, complement components, and VEGF receptors. T, B and plasma cells were polyclonal but showed class-switched and somatically hypermutated IgG1+ plasma cells consistent with stromal cell-driven germinal center activation. In conclusion, our findings show that stromal cell activation and associated B-cell activation and differentiation, neovascularization and stromal remodeling underlie CD and suggest new targets for treatment. Key pointsO_LICastleman Disease is characterized by activation and proliferation of CXCL13+ FDCs, PDGFRA+ reticular cells, and ACTA2-positive PRCs. C_LIO_LIVEGF and IL-6 from lymph node stromal cells are associated with B-cell activation and differentiation, endothelial proliferation, and inflammation in CD C_LI

pathology↗

bioRxiv2024

Mapping the Cellular Biogeography of Human Bone Marrow Niches Using Single-Cell Transcriptomics and Proteomic Imaging

The bone marrow is the organ responsible for blood production. Diverse non-hematopoietic cells contribute essentially to hematopoiesis. However, these cells and their spatial organization remain largely uncharacterized as they have been technically challenging to study in humans. Here, we used fresh femoral head samples and performed single-cell RNA sequencing (scRNA-Seq) to profile 29,325 enriched non-hematopoietic bone marrow cells and discover nine transcriptionally distinct subtypes. We next employed CO-detection by inDEXing (CODEX) multiplexed imaging of 18 individuals, including both healthy and acute myeloid leukemia (AML) samples, to spatially profile over one million single cells with a novel 53-antibody panel. We discovered a relatively hyperoxygenated arterio-endosteal niche for early myelopoiesis, and an adipocytic, but not endosteal or perivascular, niche for early hematopoietic stem and progenitor cells. We used our atlas to predict cell type labels in new bone marrow images and used these predictions to uncover mesenchymal stromal cell (MSC) expansion and leukemic blast/MSC-enriched spatial neighborhoods in AML patient samples. Our work represents the first comprehensive, spatially-resolved multiomic atlas of human bone marrow and will serve as a reference for future investigation of cellular interactions that drive hematopoiesis.

cell biology↗

bioRxiv2024

A longitudinal single-cell and spatial multiomic atlas of pediatric high-grade glioma

Pediatric high-grade glioma (pHGG) is an incurable central nervous system malignancy that is a leading cause of pediatric cancer death. While pHGG shares many similarities to adult glioma, it is increasingly recognized as a molecularly distinct, yet highly heterogeneous disease. In this study, we longitudinally profiled a molecularly diverse cohort of 16 pHGG patients before and after standard therapy through single-nucleus RNA and ATAC sequencing, whole-genome sequencing, and CODEX spatial proteomics to capture the evolution of the tumor microenvironment during progression following treatment. We found that the canonical neoplastic cell phenotypes of adult glioblastoma are insufficient to capture the range of tumor cell states in a pediatric cohort and observed differential tumor-myeloid interactions between malignant cell states. We identified key transcriptional regulators of pHGG cell states and did not observe the marked proneural to mesenchymal shift characteristic of adult glioblastoma. We showed that essential neuromodulators and the interferon response are upregulated post-therapy along with an increase in non-neoplastic oligodendrocytes. Through in vitro pharmacological perturbation, we demonstrated novel malignant cell-intrinsic targets. This multiomic atlas of longitudinal pHGG captures the key features of therapy response that support distinction from its adult counterpart and suggests therapeutic strategies which are targeted to pediatric gliomas.

cancer biology↗

bioRxiv2023

Oncogenic RAS-Pathway Activation Drives Oncofetal Reprogramming and Creates Therapeutic Vulnerabilities in Juvenile Myelomonocytic Leukemia

Aberrant fetal gene expression facilitates tumor-specific cellular plasticity by hijacking molecular programs of embryogenesis1. Persistent fetal gene signatures in childhood malignancies are typically explained by their prenatal origins2-6. In contrast, reactivation of fetal gene expression is considered a consequence of oncofetal reprogramming (OFR) in adult malignancies and is associated with aggressive disease7-10. To date, OFR has not been described in the context of childhood malignancies. Here, we performed a comprehensive multi-layered molecular characterization of juvenile myelomonocytic leukemia (JMML) and identified OFR as a hallmark of aggressive JMML. We observed that hematopoietic stem cells (HSCs) aberrantly express mixed developmental programs in JMML. Expression of fetal gene signatures combined with a postnatal epigenetic landscape suggested OFR, which was validated in a JMML mouse model, demonstrating that postnatal activation of RAS signaling is sufficient to induce fetal gene signatures. Integrative analysis identified the fetal HSC maturation marker CD52 as a novel therapeutic target for aggressive JMML. Anti-CD52 treatment depleted human JMML HSCs and disrupted disease propagation in vivo. In summary, this study implicates OFR, defined as postnatal acquisition of fetal transcription signatures, in the pathobiology of a childhood malignancy. We provide evidence for the direct involvement of oncogenic RAS signaling in OFR. Finally, we demonstrate how OFR can be leveraged for the development of novel treatment strategies. Highlights{blacksquare} Epigenomic and transcriptomic landscape of juvenile myelomonocytic leukemia (JMML) in the context of hematopoietic development. {blacksquare}The presence of fetal transcription signatures in childhood malignancies is not indicative of a developmental maturation block. {blacksquare}High-risk JMML is characterized by oncofetal reprogramming of postnatal hematopoietic stem cells (HSCs). {blacksquare}RAS-pathway mutations induce fetal-like gene expression signatures in murine postnatal HSCs. {blacksquare}The fetal maturation marker CD52 is a novel therapeutic target in high-risk JMML.

cancer biology↗

bioRxiv2023

Acetyl-CoA carboxylase 1-dependent lipogenesis drives breast cancer progression

Dysregulation of cellular energetics, including lipid synthesis mediated through de novo lipogenesis, is a feature of many cancers. Here we report that acetyl-coenzyme A carboxylase (ACC) 1, the rate-limiting enzyme of de novo lipogenesis, is a key regulator of breast cancer progression and cancer cell phenotype. Mammary epithelial-specific deletion of ACC1 impaired tumour progression and decreased cancer cell proliferation in the PyMT model of breast cancer in vivo. ACC1 knockout in human breast cancer cell lines resulted in decreased cell number and altered cell and membrane morphology. Lipidomic profiling demonstrated reduced levels of acyl-carnitines (CARs) and several phospholipid (PL) classes, whilst also shifting the lipid profiles to exhibit more elongated and less saturated lipids in ACC1 knockout breast cancer cells. Palmitate rescue of ACC1 deletion phenotypes demonstrated a critical role for ACC1 driven de novo lipogenesis in breast cancer cell function. Analysis of human breast tumour-microarrays identified strong ACC1 expression at all breast cancer stages, grade and metastasis, compared to normal adjacent tissue. Together our data demonstrate a novel role for ACC1 in breast cancer progression and cancer cell function, mediated through its lipogenic role, that together with its expression profile, identify ACC1 as a potential therapeutic target in breast cancer. Statement of significanceThis study investigates the impact of ACC1 deletion in breast cancer progression, revealing the importance of ACC1-derived lipids in breast cancer cell phenotypes and identifies ACC1 as a potential novel therapeutic target.

cancer biology↗

bioRxiv2023

Inadequate structural constraint on Fab approach rather than paratope elicitation limits HIV-1 MPER vaccine utility

Broadly neutralizing antibodies (bnAbs) against HIV-1 target conserved epitopes, thereby inhibiting viral entry. Yet surprisingly, those recognizing linear epitopes in the HIV-1 gp41 membrane proximal external region (MPER) are elicited neither by peptide nor protein scaffold vaccines. Here, we observe that while Abs generated by MPER/liposome vaccines may exhibit human bnAb-like paratopes, B-cell programming without constraints imposed by the gp160 ectodomain selects Abs unable to access the MPER within its native "crawlspace". During natural infection, the flexible hinge of IgG3 partially mitigates steric occlusion of less pliable IgG1 subclass Abs with identical MPER specificity, until affinity maturation refines entry mechanisms. The IgG3 subclass maintains B-cell competitiveness, exploiting bivalent ligation resulting from greater intramolecular Fab arm length, offsetting weak antibody affinity. These findings suggest future immunization strategies.

immunology↗

bioRxiv2023

RUNX1 is required in granulocyte-monocyte progenitors to attenuate inflammatory cytokine production by neutrophils

The transcription factor RUNX1 is mutated in familial platelet disorder with associated myeloid malignancies (FPDMM) and in sporadic myelodysplastic syndrome and leukemia. RUNX1 regulates inflammation in multiple cell types. Here we show that RUNX1 is required in granulocyte-monocyte progenitors (GMPs) to restrict the inflammatory response of neutrophils to toll-like receptor 4 (TLR4) signaling. Loss of RUNX1 in GMPs increased the TLR4 coreceptor CD14 on neutrophils, which contributed to neutrophils increased inflammatory cytokine production in response to the TLR4 ligand lipopolysaccharide. RUNX1 loss increased the chromatin accessibility of retrotransposons in GMPs and neutrophils and induced a type I interferon signature characterized by enriched footprints for signal transducer and activator of transcription (STAT1::STAT2) and interferon regulatory factors (IRF) in opened chromatin, and increased expression of interferon-stimulated genes. The overproduction of inflammatory cytokines by neutrophils was reversed by inhibitors of type I IFN signaling. We conclude that RUNX1 restrains the chromatin accessibility of retrotransposons in GMPs and neutrophils, and that loss of RUNX1 increases proinflammatory cytokine production by elevating tonic type I interferon signaling.

developmental biology↗

bioRxiv2022

Effects of breast fibroepithelial tumor associated retinoic acid receptor alpha ligand binding domain mutations on receptor function and retinoid signaling

Point mutations in the ligand binding domain of retinoic acid receptor alpha (RAR) have been implicated in breast fibroepithelial tumors development. However, their role in the tumorigenesis of solid tumors is currently unknown. In this study, using a combination of biochemical and cellular assays, we evaluated the functional consequences of known tumor associated RAR mutations on retinoic acid signaling. All of the clinically associated mutants tested showed diminished transcriptional activities compared to wild type RAR. These mutants also exhibited a dominant negative effect, an activity which has previously been linked to developmental defects and tumor formation in mice. X-ray crystallography showed that mutants remain relatively intact structurally and the loss of transcriptional activity is due to altered co-activator recruitment. In agreement with our biochemical analyses, transcriptomics and cell growth analysis showed that the mutant RAR proteins confer resistance to growth inhibition in the presence of its ligand in phyllodes tumor cells. Although the mutations impair the receptor responses to retinoic acid, certain mutant RAR are partially reactivatable with alternative synthetic agonists. Our data provide insights into the mechanisms by which RAR mutations impact tumorigenesis.

cancer biology↗

bioRxiv2022

Learning predictive models of tissue cellular neighborhoods from cell phenotypes with graph pooling

It remains poorly understood how different cell types organize and coordinate with each other to support tissue functions. We describe CytoCommunity for identification of tissue cellular neighborhoods (TCNs) based on cell phenotypes and their spatial distributions. CytoCommunity learns a mapping directly from cell phenotype space to TCN space by a graph neural network model without using additional gene or protein expression features and is thus applicable to tissue imaging data with a small number of measured features. By leveraging graph pooling, CytoCommunity enables de novo identification of condition-specific TCNs under the supervision of image labels. Using various types of single-cell-resolution spatial proteomics and transcriptomics images, we demonstrate that CytoCommunity can identify TCNs of variable sizes with substantial improvement over existing methods. To further evaluate the ability of CytoCommunity for discovering condition-specific TCNs by supervised learning, we apply it to colorectal and breast cancer tissue images with clinical outcome information. Our analysis reveals novel granulocyte- and cancer associated fibroblast-enriched TCNs specific to high-risk tumors as well as altered tumor-immune and tumor-stromal interactions within and between TCNs compared to low-risk tumors. CytoCommunity represents the first computational tool for end-to-end unsupervised and supervised analyses of single-cell spatial maps and enables direct discovery of conditional-specific cell-cell communication patterns across variable spatial scales.

bioinformatics↗