Elucidating tumor heterogeneity from spatially resolved transcriptomics data by multi-view graph collaborative learning

Single-cell omics: experimental workflow, data analyses and applications

Cells are the fundamental units of biological systems and exhibit unique development trajectories and molecular features. Our exploration of how the genomes orchestrate the formation and maintenance of each cell, and control the cellular phenotypes of various organismsis, is both captivating and intricate. Since the inception of the first single-cell RNA technology, technologies related to single-cell sequencing have experienced rapid advancements in recent years. These technologies have expanded horizontally to include single-cell genome, epigenome, proteome, and metabolome, while vertically, they have progressed to integrate multiple omics data and incorporate additional information such as spatial scRNA-seq and CRISPR screening. Single-cell omics represent a groundbreaking advancement in the biomedical field, offering profound insights into the understanding of complex diseases, including cancers. Here, we comprehensively summarize recent advances in single-cell omics technologies, with a specific focus on the methodology section. This overview aims to guide researchers in selecting appropriate methods for single-cell sequencing and related data analysis.

SpatialCVGAE: Consensus Clustering Improves Spatial Domain Identification of Spatial Transcriptomics Using VGAE

The advent of spatially resolved transcriptomics (SRT) has provided critical insights into the spatial context of tissue microenvironments. Spatial clustering is a fundamental aspect of analyzing spatial transcriptomics data. However, spatial clustering methods often suffer from instability caused by the sparsity and high noise in the SRT data. To address this challenge, we propose SpatialCVGAE, a consensus clustering framework designed for SRT data analysis. SpatialCVGAE adopts the expression of high-variable genes from different dimensions along with multiple spatial graphs as inputs to variational graph autoencoders (VGAEs), learning multiple latent representations for clustering. These clustering results are then integrated using a consensus clustering approach, which enhances the model's stability and robustness by combining multiple clustering outcomes. Experiments demonstrate that SpatialCVGAE effectively mitigates the instability typically associated with non-ensemble deep learning methods, significantly improving both the stability and accuracy of the results. Compared to previous non-ensemble methods in representation learning and post-processing, our method fully leverages the diversity of multiple representations to accurately identify spatial domains, showing superior robustness and adaptability. All code and public datasets used in this paper are available at https://github.com/wenwenmin/SpatialCVGAE .

A contrastive learning approach to integrate spatial transcriptomics and histological images

The rapid growth of spatially resolved transcriptomics technology provides new perspectives on spatial tissue architecture. Deep learning has been widely applied to derive useful representations for spatial transcriptome analysis. However, effectively integrating spatial multi-modal data remains challenging. Here, we present ConGcR, a contrastive learning-based model for integrating gene expression, spatial location, and tissue morphology for data representation and spatial tissue architecture identification. Graph convolution and ResNet were used as encoders for gene expression with spatial location and histological image inputs, respectively. We further enhanced ConGcR with a graph auto-encoder as ConGaR to better model spatially embedded representations. We validated our models using 16 human brains, four chicken hearts, eight breast tumors, and 30 human lung spatial transcriptomics samples. The results showed that our models generated more effective embeddings for obtaining tissue architectures closer to the ground truth than other methods. Overall, our models not only can improve tissue architecture identification's accuracy but also may provide valuable insights and effective data representation for other tasks in spatial transcriptome analyses.

Deep clustering representation of spatially resolved transcriptomics data using multi-view variational graph auto-encoders with consensus clustering

The rapid development of spatial transcriptomics (ST) technology has provided unprecedented opportunities to understand tissue relationships and functions within specific spatial contexts. Accurate identification of spatial domains is crucial for downstream spatial transcriptomics analysis. However, effectively combining gene expression data, histological images and spatial coordinate data to identify spatial domains remains a challenge. To this end, we propose STMVGAE, a novel spatial transcriptomics analysis tool that combines a multi-view variational graph autoencoder with a consensus clustering framework. STMVGAE begins by extracting histological images features using a pre-trained convolutional neural network (CNN) and integrates these features with gene expression data to generate augmented gene expression profiles. Subsequently, multiple graphs (views) are constructed using various similarity measures, capturing different aspects of the spatial and transcriptional relationships. These views, combined with the augmented gene expression data, are then processed through variational graph auto-encoders (VGAEs) to learn multiple low-dimensional latent embeddings. Finally, the model employs a consensus clustering method to integrate the clustering results derived from these embeddings, significantly improving clustering accuracy and stability. We applied STMVGAE to five real datasets and compared it with five state-of-the-art methods, showing that STMVGAE consistently achieves competitive results. We assessed its capabilities in spatial domain identification and evaluated its performance across various downstream tasks, including UMAP visualization, PAGA trajectory inference, spatially variable gene (SVG) identification, denoising, batch integration, and other analyses. All code and public datasets used in this paper is available at https://github.com/wenwenmin/STMVGAE and https://zenodo.org/records/13119867.

eMCI: An Explainable Multimodal Correlation Integration Model for Unveiling Spatial Transcriptomics and Intercellular Signaling

Current integration methods for single-cell RNA sequencing (scRNA-seq) data and spatial transcriptomics (ST) data are typically designed for specific tasks, such as deconvolution of cell types or spatial distribution prediction of RNA transcripts. These methods usually only offer a partial analysis of ST data, neglecting the complex relationship between spatial expression patterns underlying cell-type specificity and intercellular cross-talk. Here, we present eMCI, an explainable multimodal correlation integration model based on deep neural network framework. eMCI leverages the fusion of scRNA-seq and ST data using different spot-cell correlations to integrate multiple synthetic analysis tasks of ST data at cellular level. First, eMCI can achieve better or comparable accuracy in cell-type classification and deconvolution according to wide evaluations and comparisons with state-of-the-art methods on both simulated and real ST datasets. Second, eMCI can identify key components across spatial domains responsible for different cell types and elucidate the spatial expression patterns underlying cell-type specificity and intercellular communication, by employing an attribution algorithm to dissect the visual input. Especially, eMCI has been applied to 3 cross-species datasets, including zebrafish melanomas, soybean nodule maturation, and human embryonic lung, which accurately and efficiently estimate per-spot cell composition and infer proximal and distal cellular interactions within the spatial and temporal context. In summary, eMCI serves as an integrative analytical framework to better resolve the spatial transcriptome based on existing single-cell datasets and elucidate proximal and distal intercellular signal transduction mechanisms over spatial domains without requirement of biological prior reference. This approach is expected to facilitate the discovery of spatial expression patterns of potential biomolecules with cell type and cell-cell communication specificity.

GraphCVAE: Uncovering cell heterogeneity and therapeutic target discovery through residual and contrastive learning

Advancements in Spatial Transcriptomics (ST) technologies in recent years have transformed the analysis of tissue structure and function within spatial contexts. However, accurately identifying spatial domains remains challenging due to data sparsity and noise. Traditional clustering methods often fail to capture spatial dependencies, while spatial clustering methods struggle with batch effects and data integration. We introduce GraphCVAE, a model designed to enhance spatial domain identification by integrating spatial and morphological information, correcting batch effects, and managing heterogeneous data. GraphCVAE employs a multi-layer Graph Convolutional Network (GCN) and a variational autoencoder to improve the representation and integration of spatial information. Through contrastive learning, the model captures subtle differences between cell types and states. Extensive testing on various ST datasets demonstrates GraphCVAE's robustness and biological contributions. In the dorsolateral prefrontal cortex (DLPFC) dataset, it accurately delineates cortical layer boundaries. In glioblastoma, GraphCVAE reveals critical therapeutic targets such as TF and NFIB. In colorectal cancer, it explores the role of the extracellular matrix in colorectal cancer. The model's performance metrics consistently surpass existing methods, validating its effectiveness. GraphCVAE's advanced visualization capabilities further highlight its precision in resolving spatial structures, making it a powerful tool for spatial transcriptomics analysis and offering new insights into disease studies.

SpaGIC: graph-informed clustering in spatial transcriptomics via self-supervised contrastive learning

Spatial transcriptomics technologies enable the generation of gene expression profiles while preserving spatial context, providing the potential for in-depth understanding of spatial-specific tissue heterogeneity. Leveraging gene and spatial data effectively is fundamental to accurately identifying spatial domains in spatial transcriptomics analysis. However, many existing methods have not yet fully exploited the local neighborhood details within spatial information. To address this issue, we introduce SpaGIC, a novel graph-based deep learning framework integrating graph convolutional networks and self-supervised contrastive learning techniques. SpaGIC learns meaningful latent embeddings of spots by maximizing both edge-wise and local neighborhood-wise mutual information of graph structures, as well as minimizing the embedding distance between spatially adjacent spots. We evaluated SpaGIC on seven spatial transcriptomics datasets across various technology platforms. The experimental results demonstrated that SpaGIC consistently outperformed existing state-of-the-art methods in several tasks, such as spatial domain identification, data denoising, visualization, and trajectory inference. Additionally, SpaGIC is capable of performing joint analyses of multiple slices, further underscoring its versatility and effectiveness in spatial transcriptomics research.

Benchmarking clustering, alignment, and integration methods for spatial transcriptomics

BACKGROUND

Spatial transcriptomics (ST) is advancing our understanding of complex tissues and organisms. However, building a robust clustering algorithm to define spatially coherent regions in a single tissue slice and aligning or integrating multiple tissue slices originating from diverse sources for essential downstream analyses remains challenging. Numerous clustering, alignment, and integration methods have been specifically designed for ST data by leveraging its spatial information. The absence of comprehensive benchmark studies complicates the selection of methods and future method development.

RESULTS

In this study, we systematically benchmark a variety of state-of-the-art algorithms with a wide range of real and simulated datasets of varying sizes, technologies, species, and complexity. We analyze the strengths and weaknesses of each method using diverse quantitative and qualitative metrics and analyses, including eight metrics for spatial clustering accuracy and contiguity, uniform manifold approximation and projection visualization, layer-wise and spot-to-spot alignment accuracy, and 3D reconstruction, which are designed to assess method performance as well as data quality. The code used for evaluation is available on our GitHub. Additionally, we provide online notebook tutorials and documentation to facilitate the reproduction of all benchmarking results and to support the study of new methods and new datasets.

CONCLUSIONS

Our analyses lead to comprehensive recommendations that cover multiple aspects, helping users to select optimal tools for their specific needs and guide future method development.

Foundation models in molecular biology

Determining correlations between molecules at various levels is an important topic in molecular biology. Large language models have demonstrated a remarkable ability to capture correlations from large amounts of data in the field of natural language processing as well as image generation, and correlations captured from data using large language models can also be applicable to solving a wide range of specific tasks, hence large language models are also referred to as foundation models. The massive amount of data that exists in the field of molecular biology provides an excellent basis for the development of foundation models, and the recent emergence of foundation models in the field of molecular biology has really pushed the entire field forward. We summarize the foundation models developed based on RNA sequence data, DNA sequence data, protein sequence data, single-cell transcriptome data, and spatial transcriptome data respectively, and further discuss the research directions for the development of foundation models in molecular biology.

Dissecting tumor microenvironment from spatially resolved transcriptomics data by heterogeneous graph learning

Spatially resolved transcriptomics (SRT) has enabled precise dissection of tumor-microenvironment (TME) by analyzing its intracellular molecular networks and intercellular cell-cell communication (CCC). However, lacking computational exploration of complicated relations between cells, genes, and histological regions, severely limits the ability to interpret the complex structure of TME. Here, we introduce stKeep, a heterogeneous graph (HG) learning method that integrates multimodality and gene-gene interactions, in unraveling TME from SRT data. stKeep leverages HG to learn both cell-modules and gene-modules by incorporating features of diverse nodes including genes, cells, and histological regions, allows for identifying finer cell-states within TME and cell-state-specific gene-gene relations, respectively. Furthermore, stKeep employs HG to infer CCC for each cell, while ensuring that learned CCC patterns are comparable across different cell-states through contrastive learning. In various cancer samples, stKeep outperforms other tools in dissecting TME such as detecting bi-potent basal populations, neoplastic myoepithelial cells, and metastatic cells distributed within the tumor or leading-edge regions. Notably, stKeep identifies key transcription factors, ligands, and receptors relevant to disease progression, which are further validated by the functional and survival analysis of independent clinical data, thereby highlighting its clinical prognostic and immunotherapy applications.

Dissecting Spatiotemporal Structures in Spatial Transcriptomics via Diffusion-Based Adversarial Learning

Recent advancements in spatial transcriptomics (ST) technologies offer unprecedented opportunities to unveil the spatial heterogeneity of gene expression and cell states within tissues. Despite these capabilities of the ST data, accurately dissecting spatiotemporal structures (e.g., spatial domains, temporal trajectories, and functional interactions) remains challenging. Here, we introduce a computational framework, PearlST (partial differential equation [PDE]-enhanced adversarial graph autoencoder of ST), for accurate inference of spatiotemporal structures from the ST data using PDE-enhanced adversarial graph autoencoder. PearlST employs contrastive learning to extract histological image features, integrates a PDE-based diffusion model to enhance characterization of spatial features at domain boundaries, and learns the latent low-dimensional embeddings via Wasserstein adversarial regularized graph autoencoders. Comparative analyses across multiple ST datasets with varying resolutions demonstrate that PearlST outperforms existing methods in spatial clustering, trajectory inference, and pseudotime analysis. Furthermore, PearlST elucidates functional regulations of the latent features by linking intercellular ligand-receptor interactions to most contributing genes of the low-dimensional embeddings, as illustrated in a human breast cancer dataset. Overall, PearlST proves to be a powerful tool for extracting interpretable latent features and dissecting intricate spatiotemporal structures in ST data across various biological contexts.

A multi-view graph contrastive learning framework for deciphering spatially resolved transcriptomics data

Spatially resolved transcriptomics data are being used in a revolutionary way to decipher the spatial pattern of gene expression and the spatial architecture of cell types. Much work has been done to exploit the genomic spatial architectures of cells. Such work is based on the common assumption that gene expression profiles of spatially adjacent spots are more similar than those of more distant spots. However, related work might not consider the nonlocal spatial co-expression dependency, which can better characterize the tissue architectures. Therefore, we propose MuCoST, a Multi-view graph Contrastive learning framework for deciphering complex Spatially resolved Transcriptomic architectures with dual scale structural dependency. To achieve this, we employ spot dependency augmentation by fusing gene expression correlation and spatial location proximity, thereby enabling MuCoST to model both nonlocal spatial co-expression dependency and spatially adjacent dependency. We benchmark MuCoST on four datasets, and we compare it with other state-of-the-art spatial domain identification methods. We demonstrate that MuCoST achieves the highest accuracy on spatial domain identification from various datasets. In particular, MuCoST accurately deciphers subtle biological textures and elaborates the variation of spatially functional patterns.

Single-cell characterization of infiltrating T cells identifies novel targets for gallbladder cancer immunotherapy

Gallbladder cancer (GBC) is among the most common malignancies of biliary tract system due to its limited treatments. The immunotherapeutic targets for T cells are appealing, however, heterogeneity of T cells hinds its further development. We systematically construct T cell atlas by single-cell RNA sequencing; and utilized the identified gene signatures of high_CNV_T cells to predict molecular subtyping towards personalized therapeutic treatments for GBC. We identified 12 T cell subtypes, where exhausted CD8+ T cells, activated/exhausted CD8+ T cells, and regulatory T cells were predominant in tumors. There appeared to be an inverse relationship between Th17 and Treg populations with Th17 levels significantly reduced, whereas Tregs were concomitantly increased. Furthermore, we first established subtyping criterion to identify three subtypes of GBC based on their pro-tumorigenic microenvironments, e.g., the type 1 group shows more M2 macrophages infiltration, while the type 2 group is infiltrated by highly exhausted CD8+ T cells, B cells and Tregs with suppressive activities. Our study provides valuable insights into T cell heterogeneity and suggests that molecular subtyping based on T cells might provide a potential immunotherapeutic strategy to improve GBC treatment.

Benchmarking spatial clustering methods with spatially resolved transcriptomics data

Spatial clustering, which shares an analogy with single-cell clustering, has expanded the scope of tissue physiology studies from cell-centroid to structure-centroid with spatially resolved transcriptomics (SRT) data. Computational methods have undergone remarkable development in recent years, but a comprehensive benchmark study is still lacking. Here we present a benchmark study of 13 computational methods on 34 SRT data (7 datasets). The performance was evaluated on the basis of accuracy, spatial continuity, marker genes detection, scalability, and robustness. We found existing methods were complementary in terms of their performance and functionality, and we provide guidance for selecting appropriate methods for given scenarios. On testing additional 22 challenging datasets, we identified challenges in identifying noncontinuous spatial domains and limitations of existing methods, highlighting their inadequacies in handling recent large-scale tasks. Furthermore, with 145 simulated data, we examined the robustness of these methods against four different factors, and assessed the impact of pre- and postprocessing approaches. Our study offers a comprehensive evaluation of existing spatial clustering methods with SRT data, paving the way for future advancements in this rapidly evolving field.

MENDER: fast and scalable tissue structure identification in spatial omics data

Tissue structure identification is a crucial task in spatial omics data analysis, for which increasingly complex models, such as Graph Neural Networks and Bayesian networks, are employed. However, whether increased model complexity can effectively lead to improved performance is a notable question in the field. Inspired by the consistent observation of cellular neighborhood structures across various spatial technologies, we propose Multi-range cEll coNtext DEciphereR (MENDER), for tissue structure identification. Applied on datasets of 3 brain regions and a whole-brain atlas, MENDER, with biology-driven design, offers substantial improvements over modern complex models while automatically aligning labels across slices, despite using much less running time than the second-fastest. MENDER's identification power allows the uncovering of previously overlooked spatial domains that exhibit strong associations with brain aging. MENDER's scalability makes it freely appliable on a million-level brain spatial atlas. MENDER's discriminative power enables the differentiation of breast cancer patient subtypes obscured by single-cell analysis.

Engineering Heterogeneous Tumor Models for Biomedical Applications

Tumor tissue engineering holds great promise for replicating the physiological and behavioral characteristics of tumors in vitro. Advances in this field have led to new opportunities for studying the tumor microenvironment and exploring potential anti-cancer therapeutics. However, the main obstacle to the widespread adoption of tumor models is the poor understanding and insufficient reconstruction of tumor heterogeneity. In this review, the current progress of engineering heterogeneous tumor models is discussed. First, the major components of tumor heterogeneity are summarized, which encompasses various signaling pathways, cell proliferations, and spatial configurations. Then, contemporary approaches are elucidated in tumor engineering that are guided by fundamental principles of tumor biology, and the potential of a bottom-up approach in tumor engineering is highlighted. Additionally, the characterization approaches and biomedical applications of tumor models are discussed, emphasizing the significant role of engineered tumor models in scientific research and clinical trials. Lastly, the challenges of heterogeneous tumor models in promoting oncology research and tumor therapy are described and key directions for future research are provided.

Spatial modelling of the tumor microenvironment from multiplex immunofluorescence images: methods and applications

Spatial modelling methods have gained prominence with developments in high throughput imaging platforms. Multiplex immunofluorescence (mIF) provides the scope to examine interactions between tumor and immune compartment at single cell resolution using a panel of antibodies that can be chosen based on the cancer type or the clinical interest of the study. The markers can be used to identify the phenotypes and to examine cellular interactions at global and local scales. Several translational studies rely on key understanding of the tumor microenvironment (TME) to identify drivers of immune response in immunotherapy based clinical trials. To improve the success of ongoing trials, a number of retrospective approaches can be adopted to understand differences in response, recurrence and progression by examining the patient's TME from tissue samples obtained at baseline and at various time points along the treatment. The multiplex immunofluorescence (mIF) technique provides insight on patient specific cell populations and their relative spatial distribution as qualitative measures of a favorable treatment outcome. Spatial analysis of these images provides an understanding of the intratumoral heterogeneity and clustering among cell populations in the TME. A number of mathematical models, which establish clustering as a measure of deviation from complete spatial randomness, can be applied to the mIF images represented as spatial point patterns. These mathematical models, developed for landscape ecology and geographic information studies, can be applied to the TME after careful consideration of the tumor type (cold vs. hot) and the tumor immune landscape. The spatial modelling of mIF images can show observable engagement of T cells expressing immune checkpoint molecules and this can then be correlated with single-cell RNA sequencing data.

Decoding functional cell-cell communication events by multi-view graph learning on spatial transcriptomics

Cell-cell communication events (CEs) are mediated by multiple ligand-receptor (LR) pairs. Usually only a particular subset of CEs directly works for a specific downstream response in a particular microenvironment. We name them as functional communication events (FCEs) of the target responses. Decoding FCE-target gene relations is: important for understanding the mechanisms of many biological processes, but has been intractable due to the mixing of multiple factors and the lack of direct observations. We developed a method HoloNet for decoding FCEs using spatial transcriptomic data by integrating LR pairs, cell-type spatial distribution and downstream gene expression into a deep learning model. We modeled CEs as a multi-view network, developed an attention-based graph learning method to train the model for generating target gene expression with the CE networks, and decoded the FCEs for specific downstream genes by interpreting trained models. We applied HoloNet on three Visium datasets of breast cancer and liver cancer. The results detangled the multiple factors of FCEs by revealing how LR signals and cell types affect specific biological processes, and specified FCE-induced effects in each single cell. We conducted simulation experiments and showed that HoloNet is more reliable on LR prioritization in comparison with existing methods. HoloNet is a powerful tool to illustrate cell-cell communication landscapes and reveal vital FCEs that shape cellular phenotypes. HoloNet is available as a Python package at https://github.com/lhc17/HoloNet.

Revealing Tissue Heterogeneity and Spatial Dark Genes from Spatially Resolved Transcriptomics by Multiview Graph Networks

Spatially resolved transcriptomics (SRT) is capable of comprehensively characterizing gene expression patterns and providing an unbiased image of spatial composition. To fully understand the organizational complexity and tumor immune escape mechanism, we propose stMGATF, a multiview graph attention fusion model that integrates gene expression, histological images, spatial location, and gene association. To better extract information, stMGATF exploits SimCLRv2 for visual feature exaction and employs edge feature enhanced graph attention networks for the learning potential embedding of each view. A global attention mechanism is used to adaptively integrate 3 views to obtain low-dimensional representation. Applied to diverse SRT datasets, stMGATF is robust and outperforms other methods in detecting spatial domains and denoising data even with different resolutions and platforms. In particular, stMGATF contributes to the elucidation of tissue heterogeneity and extraction of 3-dimensional expression domains. Importantly, considering the associations between genes in tumors, stMGATF can identify the spatial dark genes ignored by traditional methods, which can be used to predict tumor-driving transcription factors and reveal tumor immune escape mechanisms, providing theoretical evidence for the development of new immunotherapeutic strategies.

Quantifying disorder one atom at a time using an interpretable graph neural network paradigm

Quantifying the level of atomic disorder within materials is critical to understanding how evolving local structural environments dictate performance and durability. Here, we leverage graph neural networks to define a physically interpretable metric for local disorder, called SODAS. This metric encodes the diversity of the local atomic configurations as a continuous spectrum between the solid and liquid phases, quantified against a distribution of thermal perturbations. We apply this methodology to four prototypical examples with varying levels of disorder: (1) grain boundaries, (2) solid-liquid interfaces, (3) polycrystalline microstructures, and (4) tensile failure/fracture. We also compare SODAS to several commonly used methods. Using elemental aluminum as a case study, we show how our paradigm can track the spatio-temporal evolution of interfaces, incorporating a mathematically defined description of the spatial boundary between order and disorder. We further show how to extract physics-preserved gradients from our continuous disorder fields, which may be used to understand and predict materials performance and failure. Overall, our framework provides a simple and generalizable pathway to quantify the relationship between complex local atomic structure and coarse-grained materials phenomena.