Abstract:Spatial transcriptomics and messenger RNA splicing encode extensive spatiotemporal information for cell states and transitions. The current lineage-inference methods either lack spatial dynamics for state transition or cannot capture different dynamics associated with multiple cell states and transition paths. Here we present spatial transition tensor (STT), a method that uses messenger RNA splicing and spatial transcriptomes through a multiscale dynamical model to characterize multistability in space. By learning a four-dimensional transition tensor and spatial-constrained random walk, STT reconstructs cell-state-specific dynamics and spatial state transitions via both short-time local tensor streamlines between cells and long-time transition paths among attractors. Benchmarking and applications of STT on several transcriptome datasets via multiple technologies on epithelial–mesenchymal transitions, blood development, spatially resolved mouse brain and chicken heart development, indicate STT's capability in recovering cell-state-specific dynamics and their associated genes not seen using existing methods. Overall, STT provides a consistent multiscale description of single-cell transcriptome data across multiple spatiotemporal scales.

What problem does this paper attempt to address?

The main problem that this paper attempts to solve is how to reconstruct and analyze multi - stable cell states and their transition paths in single - cell spatial transcriptome data. Specifically, the existing lineage inference methods either lack spatial dynamic information or are unable to capture different dynamics related to multiple cell states and transition paths. This paper proposes a new method - **Spatial Transition Tensor (STT)** - to overcome these limitations. ### Main problems 1. **Reconstruction of multi - stable cell states**: Existing methods have difficulty in capturing the multi - linear or multi - stable characteristics in complex tissues, that is, multiple stable states that exist simultaneously. 2. **Lack of spatial dynamic information**: Current methods cannot fully integrate spatial information to describe the spatio - temporal dynamics of cell state transitions. 3. **Inconsistent time scales**: The time scale of mRNA splicing is at the level of minutes to hours, while cell state transitions may span days to weeks, and existing models have difficulty dealing with this difference in time scales. 4. **Under - utilization of unspliced and spliced mRNA information**: Existing methods usually only focus on the spliced mRNA counts and ignore the important information of unspliced mRNA counts for gene regulation. ### Solutions STT solves the above problems in the following ways: - **Multi - dimensional transition tensor**: A four - dimensional transition tensor is constructed, covering cells, genes, splicing states and attractors, to characterize multi - stable characteristics. - **Spatially - constrained random walk**: Combined with the spatially - constrained random walk model, STT can reconstruct cell - state - specific dynamics as well as spatial state transitions. - **Transition paths at short and long time scales**: Through local tensor streamlines and global transition paths, STT can describe the fluctuations of cells around attractors in a short time and state transitions in a long period. - **Integration of unspliced and spliced mRNA information**: Using unspliced and spliced mRNA counts, STT can more comprehensively capture gene regulation information, thereby more accurately predicting the direction and speed of cell state transitions. ### Application examples The paper shows the application of STT on multiple datasets, including epithelial - mesenchymal transition (EMT), blood development, mouse brain development and chicken heart development, etc. For example, in the EMT experiment of A549 cell line, STT successfully identified three attractors (E, ICS, M) and revealed the importance of the intermediate transition state (ICS) of cells from the epithelial state to the mesenchymal state. In conclusion, STT provides a powerful tool that can reconstruct and analyze multi - stable cell states and their transition paths in single - cell spatial transcriptome data, filling the gaps in existing methods.

Spatial transition tensor of single cells

A transition tensor for quantifying spatial transcriptome attractors

DSTG: Deconvoluting Spatial Transcriptomics Data Through Graph-Based Artificial Intelligence

STdGCN: spatial transcriptomic cell-type deconvolution using graph convolutional networks

Dissecting Transition Cells from Single-Cell Transcriptome Data Through Multiscale Stochastic Dynamics.

Abstract 2129: Spatial Charting of Single Cell Transcriptomes in Tissues

Inference of Spatiotemporal Effects on Cellular State Transitions from Time-Lapse Microscopy

Deciphering tissue structure and function using spatial transcriptomics

stDiff: a diffusion model for imputing spatial transcriptomics through single-cell transcriptomics

SPACEL: Characterizing Spatial Transcriptome Architectures by Deep-Learning

SPACEL: deep learning-based characterization of spatial transcriptome architectures

stLearn: integrating spatial location, tissue morphology and gene expression to find cell types, cell-cell interactions and spatial trajectories within undissociated tissues

STEM enables mapping of single-cell and spatial transcriptomics data with transfer learning

Spatial transcriptomics in human biomedical research and clinical application

Spatial charting of single-cell transcriptomes in tissues

TIST: Transcriptome and Histopathological Image Integrative Analysis for Spatial Transcriptomics

Spatial transcriptomics in development and disease

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

stVCR: Reconstructing spatio-temporal dynamics of cell development using optimal transport

STEP: Deciphering Spatial Atlas at Single-Cell Level with Whole-Transcriptome Coverage

Dissecting reversible and irreversible single cell state transitions from gene regulatory networks