Lei Cao

Understanding complex biological systems requires tracing cellular dynamic changes across conditions, time, and space. However, integrating multi-sample data in a unified way to explore cellular heterogeneity remains challenging. Here, we present Stereopy, a flexible framework for modeling and dissecting comparative and spatiotemporal patterns in multi-sample spatial transcriptomics with interactive data visualization. To optimize this framework, we devise a universal container, a scope controller, and an integrative transformer tailored for multi-sample multimodal data storage, management, and processing. Stereopy showcases three representative applications: investigating specific cell communities and genes responsible for pathological changes, detecting spatiotemporal gene patterns by considering spatial and temporal features, and inferring three-dimensional niche-based cell-gene interaction network that bridges intercellular communications and intracellular regulations. Stereopy serves as both a comprehensive bioinformatics toolbox and an extensible framework that empowers researchers with enhanced data interpretation abilities and new perspectives for mining multi-sample spatial transcriptomics data. Tracing cellular changes in complex biological systems is challenging. Here, authors present a flexible framework that integrates multi-sample data with in-house algorithms to infer comparative and spatiotemporal cell-gene patterns, advancing understanding of cellular dynamics.

BACKGROUND: Cell clustering is a pivotal aspect of spatial transcriptomics (ST) data analysis as it forms the foundation for subsequent data mining. Recent advances in spatial domain identification have leveraged graph neural network (GNN) approaches in conjunction with spatial transcriptomics data. However, such GNN-based methods suffer from representation collapse, wherein all spatial spots are projected onto a singular representation. Consequently, the discriminative capability of individual representation feature is limited, leading to suboptimal clustering performance. RESULTS: To address this issue, we proposed SGAE, a novel framework for spatial domain identification, incorporating the power of the Siamese graph autoencoder. SGAE mitigates the information correlation at both sample and feature levels, thus improving the representation discrimination. We adapted this framework to ST analysis by constructing a graph based on both gene expression and spatial information. SGAE outperformed alternative methods by its effectiveness in capturing spatial patterns and generating high-quality clusters, as evaluated by the Adjusted Rand Index, Normalized Mutual Information, and Fowlkes-Mallows Index. Moreover, the clustering results derived from SGAE can be further utilized in the identification of 3-dimensional (3D) Drosophila embryonic structure with enhanced accuracy. CONCLUSIONS: Benchmarking results from various ST datasets generated by diverse platforms demonstrate compelling evidence for the effectiveness of SGAE against other ST clustering methods. Specifically, SGAE exhibits potential for extension and application on multislice 3D reconstruction and tissue structure investigation. The source code and a collection of spatial clustering results can be accessed at https://github.com/STOmics/SGAE/.

This paper introduces a new approach to cell clustering using the Variable Neighborhood Search (VNS) metaheuristic. The purpose of this method is to cluster cells based on both gene expression and spatial coordinates. Initially, we confronted this clustering challenge as an Integer Linear Programming minimization problem. Our approach introduced a novel model based on the VNS technique, demonstrating the efficacy in navigating the complexities of cell clustering. Notably, our method extends beyond conventional cell-type clustering to spatial domain clustering. This adaptability enables our algorithm to orchestrate clusters based on information gleaned from gene expression matrices and spatial coordinates. Our validation showed the superior performance of our method when compared to existing techniques. Our approach advances current clustering methodologies and can potentially be applied to several fields, from biomedical research to spatial data analysis.

Abstract Single-cell RNA sequencing (scRNA-seq) enables characterization of cellular heterogeneity but lacks spatial context, while Spatially Transcriptomics maps gene expression in tissues with limited single-cell resolution. Integrating the complementary strengths of these data into a unified framework remains challenging. Here, we present OmniCell, a foundation model for single-cell and spatial transcriptomics, pretrained on a large-scale corpus of 67 million single-cell and spatial transcriptomic profiles, enabling the unified multi-omics representation learning. As the first foundation model to jointly capture intra-cellular gene expression relationships and inter-cellular spatial dependencies within a unified framework, OmniCell explicitly represents tissue spatial topology by serializing spatially adjacent cells during input construction. Leveraging this unified modeling paradigm, OmniCell generates unified representations of genes, cells, and tissue spatial organization. In zero-shot evaluations, it reliably recovers cell-type structure and gene expression patterns, reconstructs co-expression relationships, and outperforms existing methods across all evaluated tasks, including cell-type deconvolution and spatial domain delineation. Applied to real spatial datasets, OmniCell resolves transitional zones at tumor margins and reveals associated inflammatory activation and immune-cell enrichment, demonstrating its capacity for high-resolution spatial profiling.

Summary The human fetal brain undergoes morphological changes that contribute to the development of regional functionalities. However, the features of structural development, the underlying molecular and cellular signatures in the fetal brain remain unclear. With spatial transcriptomics and snRNA-seq, we identified 25 forebrain regions and characterized the dynamic changes in the cortex and striatum during the late first and early second trimesters. In particular, we discovered that temporal lobe enriched NPY-expressing L2/3 EX neuron potentially interacted with L4 EX neurons during cortical expansion and arealization. Additionally, the gyrus and sulcus were developmental asynchronous, in which HOPX and SPARC genes were potentially involved. Further investigation on the striatum showed specific genes and cell types that enriched in patch and matrix compartments, and SST -positive interneurons potentially involved in the development of these structures. Together, our results give insights into the understanding of early fetal brain development.