Biology subjects

Nguyen, Q.

Publications and source records attributed to Nguyen, Q..

At least 37 records · Page 2Linked to original sources

bioRxiv2025

VN1K: a genome graph-based and function-driven multi-omics and phenomics resource for the Vietnamese population

Vietnam, the 16th most populated nation, remains profoundly underrepresented in global genomic databases. Here, we present VN1K, a first-ever comprehensive and well-curated resource of multi-omics data with a wide-range of phenotypic information of 1,011 unrelated Vietnamese individuals. High-depth short-read whole-genome sequencing data were generated for all samples along with various - omic data, including microarray, long-read whole-genome sequencing, and RNA sequencing. Using a high-sensitivity variant detection pipeline, which included a pangenome graph reference and a deep-learning framework, we identified nearly 40 million variants of which 8.5 million are novel with nearly 900 thousand short insertions/deletions and 39 thousand structural variants. Specifically, VN1K featured a first-ever whole-genome methylation profile based on long read sequencing. A genotype imputation panel was also created with the highest accuracy on the Vietnamese population. Variants with significantly different allele frequencies in the Vietnamese population compared to others were found to be functionally significant, especially in genes associated with immune diseases (HLA-B, KIR3DL3, KIR2DL1, KIR2DL4) or drug responses (CYP2C19, CYP2D6, VKORC1, CYP2B6). We were also able to map various loci related to hepatitis B virus infection as well as six disease traits, including triglyceride levels, LDL-C, serum glucose levels, HbA1c, and levels of two liver enzymes (ALT and AST). VN1K dataset is accessible via genome.vinbigdata.org, an integrated platform with both linear and graph-based genome browser for facilitating data exploration, research, and applications in precision medicine.

genomics↗

bioRxiv2025

Spatial Transcriptomic Signature of Progressive Fibrosis in Human MASLD: Role of Senescence and Metabolic Reprogramming.

Granular detail about the location and nature of liver cell interactions and the metabolic, inflammatory and fibrogenic pathways driving progressive fibrosis in metabolic dysfunction-associated steatotic liver disease (MASLD) is needed to deliver novel therapeutic targets. Here we used spatial transcriptomic data from human MASLD liver biopsies to identify the major cell types and their potential interconnected activities within specific tissue regions across the spectrum of MASLD. Gene expression data were generated using 10X Genomics Visium technology from 33 formalin-fixed paraffin-embedded liver biopsy samples and overlaid with annotated anatomical regions. Differential gene expression (DEG) and pathway analyses, cellular deconvolution and ligand-receptor interactions were conducted for each annotated anatomical category, with specific protein expression validated using CODEX spatial proteomics and immunohistochemistry staining. Unsupervised gene expression data grouped the annotated spots into 2 main clusters enriched for early/intermediate vs late fibrosis and transcriptome-based cellular deconvolution was well aligned with annotated histopathological features. In addition to extracellular matrix/receptor interactions and immune cell recruitment and trafficking, several genes encoding immunoglobulins were highly upregulated in late-stage fibrosis and were spatially associated with a senescence signature. Upregulated DEGs for early/intermediate-stage fibrosis were significantly enriched for metabolic pathways, oxidative phosphorylation and fatty acid metabolism. In contrast glycolysis genes were strongly co-expressed with late stage fibrosis. MASLD progression is accompanied by a decline in normal liver metabolic function and significant reprogramming of metabolic fuel utilisation. The spatial association of a senescence signature with expression of genes encoding immunoglobulins and complement has been linked to aging and is associated with progressive fibrosis. This work provides a valuable discovery dataset spanning different stages of human liver fibrosis and highlights the complex crosstalk between metabolic perturbations and inflammation underpinning fibrosis progression.

genomics↗

bioRxiv2025

CD8+ T cells mediate vaccination-induced lymphatic containment of latent Mycobacterium tuberculosis infection following immunosuppression, while B cells are dispensable

It is estimated that two billion people are latently infected with Mycobacterium tuberculosis (Mtb), the causative agent of tuberculosis (TB). Latent Mtb infection (LTBI) can occur in multiple organs, including the lymphatics. The risk of LTBI reactivation increases in immunocompromised conditions, such as coinfection with human immunodeficiency virus (HIV), and during treatment of autoimmune diseases and organ transplantation. The immunological correlates of protection against TB, including against reactivation of LTBI, remain largely elusive. Here, we used a mouse model of latent lymphatic Mtb infection to dissect the immunological mechanisms underlying LTBI containment versus reactivation. We show that immunosuppression-mediated reactivation of lymphatic LTBI and the subsequent spread to non-lymphatic organs can be prevented by vaccination with multiple recombinant BCG (rBCG) strains despite the deficiency of CD4+ T cells. Using spatial transcriptomics, multi-parameter imaging, network analysis and bioinformatic integration of histopathological images, we reveal that immunosuppression is associated with a distinct repositioning of non-CD4 immune cells at the edge of TB lesions within the infection-draining cervical lymph nodes. While B cells increased in numbers, they are dispensable for the containment of LTBI. Lymphatic Mtb infection in different immune cell-deficient mouse strains, antibody-mediated cell depletion and adoptive transfer experiments into highly susceptible mice unequivocally show that vaccination-mediated prevention of LTBI reactivation is critically dependent on CD8+ T cells. These findings have profound implications for our understanding of immunity to TB and the management of LTBI. Graphical Abstract O_FIG O_LINKSMALLFIG WIDTH=200 HEIGHT=92 SRC="FIGDIR/small/634479v1_ufig1.gif" ALT="Figure 1"> View larger version (26K): org.highwire.dtl.DTLVardef@151fdd6org.highwire.dtl.DTLVardef@1efed47org.highwire.dtl.DTLVardef@c2e7e0org.highwire.dtl.DTLVardef@555c7_HPS_FORMAT_FIGEXP M_FIG C_FIG

immunology↗

bioRxiv2024

Reconstructing 3D Spatial Structures of RNA-Tomography Transcriptomes by Tensor Factorization

Cells are organized to form three-dimensional structures of complex tissues. To map the complete 3D organization of a tissue, technologies based on tissue microdissections provide deep bulk RNA sequencing of orthographically arranged cryosections of a tissue, such that the full 3D spatial structure could be inferred from deeply sequenced transcriptomes in three views projected similarly as 3D tomography. Here, we introduce CTFacTomo to learn a Collapsed Tensor Factorization for RNA tomography data from cryosections to reconstruct 3D spatially resolved gene expressions. CTFacTomo combines tensor factorization with collapsing tensor entries to match the bulk gene expressions in each cryosection, enriched by a regularization of a product graph of protein-protein interaction network and spatial graphs. In the experiments, CTFacTomo is first validated on three datasets projected from fully profiled 3D spatial gene expressions to demonstrate that CTFacTomo significantly outperforms the benchmark methods for predicting the ground-truth gene expressions based on the projected 1D spatial gene expressions of three orthographic views. CTFacTomo is then applied to two RNA tomography datasets from zebrafish embryo and mouse olfactory mucosa, respectively. In both datasets, CTFacTomo detects 3D spatial expressions of several marker genes that are consistent with the developmental or functional regions in comparison to accompanying ISH staining images. In addition, a qualitative comparison between the reconstructed zebrafish embryo gene expressions with a matched external 3D Stereo-seq dataset also suggests that CTFacTomo reconstructs more spatially coherent patterns in the whole transcriptome with state-of-the-art performance.

bioinformatics↗

bioRxiv2024

The single-cell spatial landscape of stage III colorectal cancers

We conducted a spatial analysis using imaging mass cytometry applied to stage III colorectal adenocarcinomas. This study used multiplexed markers to distinguish individual cells and their spatial organization from 52 colorectal cancers. We determined the landscape features of cellular spatial features in the CRC tumor microenvironment. This spatial single-cell analysis identified 10 unique cell phenotypes in the tumor microenvironment that included stromal and immune cells with a subset which had a proliferative phenotype. These special features included spatial neighborhood interactions between single cells as well as different tissue niches, especially the tumor infiltrating lymphocyte regions. We applied a robust statistical analysis to identify significant correlations of cell features with phenotypes such as microsatellite instability or recurrence. We determined that microsatellite stable (MSS) colorectal cancers had an increased risk of recurrence if they had the following features: 1) a low level of stromal tumor-infiltrating lymphocytes, and 2) low interactions between CD4+ T cells and stromal cells. Our results point to the utility of spatial single-cell interaction analysis in defining novel features of the tumor immune microenvironments and providing useful clinical cell-related spatial biomarkers.

cancer biology↗

bioRxiv2024

SpaMTP: Integrative Statistical Analysis and Visualisation of Spatial Metabolomics and Transcriptomics data.

The ability to spatially measure multi-modal data provides an unprecedented opportunity to comprehensively explore molecular regulation at transcriptional, translational and metabolic levels to acquire insights on cellular activities underpinning health and disease. However, there is currently a lack of analytical tools to integrate complementary information across different spatial-omics data modalities, particularly with respect to spatial metabolomics data, which is becoming increasingly invaluable. We introduce SpaMTP, a versatile software that implements an end-to-end integrative analysis of spatial metabolomics and transcriptomics data. Based in R, SpaMTP bridges processing functionalities for metabolomics data from Cardinal with user-friendly cell-centric analyses implemented in Seurat. Furthermore, SpaMTPs comprehensive analysis pipeline covers (1) automated mass-to-charge ratio (m/z) metabolite annotation; (2) a wide range of metabolite-gene based downstream statistical analyses including differential expression, pathway analysis, and correlation analysis; (3) integrative spatial-omics analysis; and (4) a suite of visualisation functions. For flexibility and interoperability, SpaMTP includes various functions for data import/export and object conversion, enabling seamless integration with other R and Python packages. We demonstrated the utility of SpaMTP to draw new biological understandings through analysing two biological system. We believe this software and implemented methods will be broadly utilised in spatial multi-omics and spatial metabolomics analyses.

bioinformatics↗

bioRxiv2024

Spatial analysis of HPV associated cervical intraepithelial neoplastic tissues demonstrate distinct immune signatures associated with cervical cancer progression

Cervical cancer remains the fourth most common cancer affecting women worldwide, and incidences of other HPV-related cancers continue to rise. For the development of effective prevention strategies in high-risk patients, we aimed to better understand the roles of inflammatory pathways and the tumour microenvironment as the main driver of progression to malignancy in HPV-infected tissues. We analysed the spatial organisation of seven samples of HPV+ high-grade squamous intraepithelial lesion (HSIL) and cervical intraepithelial neoplasia 3 (CIN3), comparing tumour heterogeneity and immune microenvironments between pre-malignant (neoplastic) and adjacent cervical tissues. We observed evidence of immune suppression within the neoplastic regions across all samples and identified distinct immune clusters for each dysplastic lesion. Previous single-cell data analyses in an HPV16 E7 oncoprotein-driven transgenic mouse model suggested a potential role for IL34-CSF1R signalling in immune modulation, where low IL34 expression was associated with Langerhans cell dysfunction, and, in cervical cancer, with poor patient outcome. Here, we observed that IL34-CSF1R co-expression was absent within HPV-associated neoplastic tissues but present in adjacent normal tissue regions. Additionally, we identified enrichment of an M2 gene signature in neoplastic tissue, while adjacent tissue was enriched with a pro-inflammatory M1 gene signature. Our findings provide bio-pathological insights into the spatial cellular and molecular mechanisms underlying HPV-associated cervical cancer immune regulation and suggest a strategy to modulate the immune system in HPV-positive neoplastic cervical and other tissues.

genomics↗

bioRxiv2024

Single-cell and spatial transcriptomics of vulvar lichen sclerosus reveal multicompartmental alterations in gene expression and signaling cross-talk

Vulvar diseases are a critical yet often neglected area of womens health, profoundly affecting patients quality of life and frequently resulting in long-term physical and psychological challenges. Lichen sclerosus (LS) is a chronic inflammatory skin disorder that predominantly affects the vulva, leading to severe itching, pain, scarring, and an increased risk of malignancy. Despite its profound impact on affected individuals, the molecular pathogenesis of vulvar LS (VLS) is not well understood, hindering the development of FDA-approved therapies. Here, we utilize single-cell and spatial transcriptomics to analyze lesional and non-lesional skin from VLS patients, as well as healthy control vulvar skin. Our findings demonstrate histologic, cellular, and molecular heterogeneities within VLS, yet highlight unifying molecular changes across keratinocytes, fibroblasts, immune cells, and melanocytes in lesional skin. They reveal cellular stress and damage in fibroblasts and keratinocytes, enhanced T cell activation and cytotoxicity, aberrant cell-cell signaling, and increased activation of the IFN, JAK/STAT, and p53 pathways in specific cell types. Using both monolayer and organotypic culture models, we also demonstrate that knockdown of select genes, which are downregulated in VLS lesional keratinocytes, partially recapitulates VLS-like stress-associated changes. Collectively, these data provide novel insights into the pathogenesis of VLS, identifying potential biomarkers and therapeutic targets for future research.

genomics↗

bioRxiv2024

Unraveling lncRNA Diversity at a Single Cell Resolution and in a Spatial Context across Different Cancer Types

Long non-coding RNAs (lncRNAs) play pivotal roles in gene regulation and disease, including cancer. Overcoming the limitations of lncRNA analysis with bulk data, we analyzed single-cell and spatial transcriptomics data to uncover 354937 novel lncRNAs and their functions across 13 cancer types. LncRNA functions were assessed by identifying their cell-type specificity and distinct spatial distributions across different tissue regions. First, lncRNAs were computationally validated by comparing to existing databases, and experimentally validated using spatial long read sequencing methods. Further, genome-wide computation of spatial-autocorrelation identified coexpression of lncRNAs with cancer-associated protein coding genes across the tissue. Additionally, genomic co-localization of lncRNAs with regulatory features and disease-associated genetic variants suggest possible functional association. The identified lncRNAs were analyzed for responses to immunotherapy and prognostic value, revealing cancer-outcome associated lncRNAs. We have made this novel resource available as an open website SPanC-Lnc hosted on AWS cloud to serve as a pan-cancer atlas of single cell- and spatially-resolved lncRNAs. These can complement established biomarkers because they reflect the unique characteristics of specific cell populations within tumors, offering new insights into disease progression and treatment response.

cancer biology↗

bioRxiv2024

Sequential emergence and contraction of epithelial subtypes in the prenatal human choroid plexus revealed by a stem cell model

Despite the major roles of choroid plexus epithelial cells (CPECs) in brain homeostasis and repair, their developmental lineage and diversity remain undefined. In simplified differentiations from human pluripotent stem cells, derived CPECs (dCPECs) displayed canonical properties and dynamic multiciliated phenotypes that interacted with A{beta} uptake. Single dCPEC transcriptomes over time correlated well with human organoid and fetal CPECs, while pseudotemporal and cell cycle analyses highlighted the direct CPEC origin from neuroepithelial cells. In addition, time series analyses defined metabolic (type 1) and ciliogenic dCPECs (type 2) at early timepoints, followed by type 1 diversification into anabolic-secretory (type 1a) and catabolic-absorptive subtypes (type 1b) as type 2 cells contracted. These temporal patterns were then confirmed in independent derivations and mapped to prenatal stages using human tissues. In addition to defining the prenatal lineage of human CPECs, these findings suggest new dynamic models of ChP support for the developing human brain.

neuroscience↗

bioRxiv2024

Spatial Transcriptomic Sequencing of a DIPG-infiltrated Brainstem reveals Key Invasion Markers and Novel Ligand-Receptor Pairs contributing to Tumour to TME Crosstalk

Emerging spatially-resolved sequencing technologies offer unprecedented possibilities to study cellular functionality and organisation, transforming our understanding of health and disease. The necessity to understand healthy and diseased tissues in its entirety becomes even more evident for the human brain, the most complex organ in the body. The brains cellular architecture and corresponding functions are tightly regulated. However, when intercellular communications are altered by pathologies, such as brain cancer, these microenvironmental interactions are disrupted. DIPG is a brainstem high-grade glioma arising in young children and is universally fatal. Major disease obstacles include intratumoural genetic and cellular heterogeneity as well as a highly invasive phenotype. Recent breakthrough studies have highlighted the vital oncogenic capacity of brain cancer cells to functionally interact with the central nervous system (CNS). This CNS-crosstalk crucially contributes to tumour cell invasion and disease progression. Ongoing worldwide efforts seek to better understand these cancer-promoting CNS interactions to develop more effective DIPG anti-cancer therapies. In this study, we performed spatial transcriptomic analysis of a complete tumour-infiltrated brainstem from a single DIPG patient. Gene signatures from ten sequential tumour regions were analysed to assess disease progression and to study DIPG cell interactions with the tumour microenvironment (TME). We leveraged this unique DIPG dataset to evaluate genes significantly correlated with invasive tumour distal regions versus the proximal tumour initiation site. Furthermore, we assessed novel ligand-receptor pairs that actively promote DIPG tumour progression via crosstalk with endothelial, neuronal and immune cell communities, which can be utilised to support future research efforts in this area of high unmet need.

cancer biology↗

bioRxiv2024

Metabolic immaturity of newborns and breast milk bile acids are the central determinants of heightened neonatal vulnerability to norovirus diarrhea

Noroviruses are the leading global cause of acute gastroenteritis, responsible for 685 million annual cases. While all age groups are susceptible to noroviruses, children are vulnerable to more severe infections than adults, underscored by 200 million pediatric cases and up to 200,000 deaths in children annually. Understanding the basis for the increased vulnerability of young hosts is critical to developing effective treatments. The pathogenic outcome of any enteric virus infection is governed by a complex interplay between the virus, intestinal microbiota, and host immune factors. A central mediator in these complex relationships are host- and microbiota-derived metabolites. Noroviruses bind a specific class of metabolites, bile acids, which are produced by the host and then modified by commensal bacterial enzymes. Paradoxically, bile acids can have both proviral and antiviral roles during norovirus infections. Considering these opposing effects, the microbiota-regulated balance of the bile acid pool may be a key determinant of the pathogenic outcome of a norovirus infection. The bile acid pool in newborns is unique due to immaturity of host metabolic pathways and developing gut microbiota, which could underlie the vulnerability of these hosts to severe norovirus infections. Supporting this concept, we demonstrate herein that microbiota and their bile acid metabolites protect from severe norovirus diarrhea whereas host-derived bile acids promote disease. Remarkably, we also report that maternal bile acid metabolism determines neonatal susceptibility to norovirus diarrhea during breastfeeding by delivering proviral bile acids to the newborn. Finally, directed targeting of maternal and neonatal bile acid metabolism can protect the neonatal host from norovirus disease. Altogether, these data support the conclusion that metabolic immaturity in newborns and ingestion of proviral maternal metabolites in breast milk are the central determinants of heightened neonatal vulnerability to norovirus disease.

microbiology↗

bioRxiv2024

MMCCI: Multimodal Cell-Cell Interaction Integrative Analysis of Single Cell and Spatial Data

Cell-cell interaction (CCI) analyses are an indispensable tool for harnessing the detail and depth of spatial and single-cell transcriptomics datasets by inferring inter-cellular communications, but no methods to integrate CCI results across samples exist currently. To address this, we have developed a computational pipeline, Multimodal CCI (MMCCI), to statistically integrate and analyze CCI results from existing popular CCI tools. We benchmarked MMCCIs integration on single-cell spatial datasets and found it to be highly accurate compared to simpler methods. We utilized MMCCIs integration and downstream biological analyses to uncover global and differential interaction patterns in multimodal aging brain and melanoma spatial datasets.

bioinformatics↗

bioRxiv2024

Named Entity Recognition of Pharmacokinetic parameters in the scientific literature

The development of accurate predictions for a new drugs absorption, distribution, metabolism, and excretion profiles in the early stages of drug development is crucial due to high candidate failure rates. The absence of comprehensive, standardised, and updated pharmacokinetic (PK) repositories limits pre-clinical predictions and often requires searching through the scientific literature for PK parameter estimates from similar compounds. While text mining offers promising advancements in automatic PK parameter extraction, accurate Named Entity Recognition (NER) of PK terms remains a bottleneck due to limited resources. This work addresses this gap by introducing novel corpora and language models specifically designed for effective NER of PK parameters. Leveraging active learning approaches, we developed an annotated corpus containing over 4,000 entity mentions found across the PK literature on PubMed. To identify the most effective model for PK NER, we fine-tuned and evaluated different NER architectures on our corpus. Fine-tuning BioBERT exhibited the best results, achieving a strict F1 score of 90.37% in recognising PK parameter mentions, significantly outperforming heuristic approaches and models trained on existing corpora. To accelerate the development of end-to-end PK information extraction pipelines and improve pre-clinical PK predictions, the PK NER models and the labelled corpus were released open source at https://github.com/PKPDAI/PKNER.

pharmacology and toxicology↗

bioRxiv2023

Pharmacological and behavioral effects of tryptamines present in psilocybin-containing mushrooms

Demand for more efficacious antidepressants, particularly those with a rapid onset of action, has resulted in a reevaluation of psychedelic drugs for their therapeutic potential. Several tryptamines found in psilocybin-containing magic mushrooms share chemical similarities with psilocybin, and early work suggests they may also share receptor targets. However, few studies have explored their pharmacological and behavioral effects. To accomplish this, we compared baeocystin, norbaeocystin, and aeruginascin with psilocybin to determine if they are metabolized by the same enzymes, penetrate the blood brain barrier, serve as ligands for similar centrally located receptors, and modulate behavior in rodents similarly. We first assessed the stability and optimal storage and handling conditions for each compound. In vitro enzyme kinetics assays then found that all compounds shared nearly identical rates of dephosphorylation via alkaline phosphatase and metabolism by monoamine oxidase. Further, we found that only the dephosphorylated products of baeocystin and norbaeocystin could cross a blood brain barrier mimetic to a similar degree as the dephosphorylated form of psilocybin, psilocin. Behaviorally, only psilocybin was found to induce head twitch responses in rats, a marker of 5HT2A agonism and indicator of the compounds hallucinogenic potential. However, like psilocybin, norbaeocystin was also found to improve outcomes in the forced swim test. All compounds were found to cause minimal changes to metrics of renal and hepatic health, suggesting innocuous safety profiles. Collectively, this work suggests that other naturally-occurring tryptamines, especially norbaeocystin, may share the same therapeutic potential as psilocybin, but without causing hallucinations. HIGHLIGHTSO_LIBaeocystin, norbaeocystin, and aeruginascin may have similar therapeutic value to psilocybin, but are understudied C_LIO_LICompound stability varied widely, with dephosphorylated forms showing lowest stability C_LIO_LIRates of metabolism by alkaline phosphatase and monoamine oxidase were similar across compounds C_LIO_LIBlood brain barrier penetration was limited to dephosphorylated forms of psilocybin, baeocystin, and norbaeocystin C_LIO_LIRat behavioral testing suggested norbaeocystin may have therapeutic utility similar to psilocybin, without causing hallucinations C_LI

pharmacology and toxicology↗

bioRxiv2023

A spatially mapped gene expression signature for intestinal stem-like cells identifies high-risk precursors of gastric cancer

ObjectiveGastric intestinal metaplasia (GIM) is a precancerous lesion that increases gastric cancer (GC) risk. The Operative Link on GIM (OLGIM) is a combined clinical-histopathologic system to risk-stratify patients with GIM. The identification of molecular biomarkers that are indicators for advanced OLGIM lesions may improve cancer prevention efforts. MethodsThis study was based on clinical and genomic data from four cohorts: 1) GAPS, a GIM cohort with detailed OLGIM severity scoring (N=303 samples); 2) the Cancer Genome Atlas (N=198); 3) a collation of in-house and publicly available scRNA-seq data (N=40), and 4) a spatial validation cohort (N=5) consisting of annotated histology slides of patients with either GC or advanced GIM. We used a multi-omics pipeline to identify, validate and sequentially parse a highly-refined signature of 26 genes which characterize high-risk GIM. ResultsUsing standard RNA-seq, we analyzed two separate, non-overlapping discovery (N=88) and validation (N=215) sets of GIM. In the discovery phase, we identified 105 upregulated genes specific for high-risk GIM (defined as OLGIM III-IV), of which 100 genes were independently confirmed in the validation set. Spatial transcriptomic profiling revealed 36 of these 100 genes to be expressed in metaplastic foci in GIM. Comparison with bulk GC sequencing data revealed 26 of these genes to be expressed in intestinal-type GC. Single-cell profiling resolved the 26-gene signature to both mature intestinal lineages (goblet cells, enterocytes) and immature intestinal lineages (stem-like cells). A subset of these genes was further validated using single-molecule multiplex fluorescence in situ hybridization. We found certain genes (TFF3 and ANPEP) to mark differentiated intestinal lineages, whereas others (OLFM4 and CPS1) localized to immature cells in the isthmic/crypt region of metaplastic glands, consistent with the findings from scRNAseq analysis. Conclusionsusing an integrated multi-omics approach, we identified a novel 26-gene expression signature for high-OLGIM precursors at increased risk for GC. We found this signature localizes to aberrant intestinal stem-like cells within the metaplastic microenvironment. These findings hold important translational significance for future prevention and early detection efforts.

cancer biology↗

bioRxiv2023

Generalization of deep learning models for predicting spatial gene expression profiles using histology images: A breast cancer case study

Spatial transcriptomics is a breakthrough technology that enables spatially-resolved measurement of molecular profiles in tissues, opening the opportunity for integrated analyses of morphology and transcriptional profiles through paired imaging and gene expression data. However, the high cost of generating data has limited its widespread adoption. Predicting gene expression profiles from histology images only can be an effective and cost-efficient in-silico spatial transcriptomics solution but is computationally challenging and current methods are limited in model performance. To advance research in this emerging and important field, this study makes the following contributions. We first provide a systematic review of deep learning methods for predicting gene expression profiles from histology images, highlighting similarities and differences in algorithm, model architecture, and data processing pipelines. Second, we performed extensive experiments to evaluate the generalization performance of the reviewed methods on several spatial transcriptomics datasets for breast cancer, where the datasets are generated using different technologies. Lastly, we propose several ideas for model improvement and empirically investigate their effectiveness. Our results shed insight on key features in a neural network model that either improve or not the performance of in-silico spatial transcriptomics, and we highlight challenges in developing algorithms with strong generalization performance. Key MessagesO_LIWe comprehensively compared the performance of existing methods for predicting spatial gene expression profiles from histology images C_LIO_LIWe assessed the roles of different algorithms, model architectures, and data processing pipelines to model performance C_LIO_LIWe performed extensive experiments to evaluate the generalization of the models on in-distribution and out-of-distribution spatial transcriptomics datasets C_LIO_LIWe proposed several strategies for improving existing models and empirically investigated their effectiveness C_LI

bioinformatics↗

bioRxiv2023

Genome-wide transcriptome analysis reveals the diversity and function of long non-coding RNAs in dinoflagellates

Dinoflagellates are a diverse group of phytoplankton, ranging from harmful bloom-forming microalgae to photosymbionts that are critical for sustaining coral reefs. Genome and transcriptome data from dinoflagellates are revealing extensive genomic divergence and lineage-specific innovation of gene functions. However, most studies thus far have focused on protein-coding genes; long non-coding RNAs (lncRNAs), known to regulate gene expression in eukaryotes, are largely unexplored. Here, using both genome and transcriptome data, we identified a combined total of 48,039 polyadenylated lncRNAs in the genomes of three dinoflagellate species: the coral symbionts of Cladocopium proliferum and Durusdinium trenchii, and the bloom-forming Prorocentrum cordatum. These putative lncRNAs are shorter, and have fewer introns and lower G+C-content when compared to protein-coding sequences. Although 37,768 (78.6%) lncRNAs shared no significant similarity with one another, we classified all lncRNAs based on conserved sequence motifs (k-mers) into distinct clusters following properties of potential protein-binding and/or subcellular localisation. Interestingly, 3708 (7.7%) lncRNAs were differentially expressed in response to heat stress, lifestyle, and/or growth phases, and they shared co-expression patterns with protein-coding genes. Based on inferred triplex interactions between lncRNA and upstream (putative promoter) regions of protein-coding genes, we identified a combined 19,460 putative gene targets for 3,721 lncRNAs; 907 genes exhibit differential expression under heat stress. These results reveal for the first time the functional diversity of lncRNAs in dinoflagellates, and demonstrate how lncRNAs, often overlooked in transcriptome data, could regulate gene expression as a molecular response to heat stress in these ecologically important organisms.

genomics↗