Geometric-aware deep learning for deciphering tissue structure from spatially resolved transcriptomics
- 1School of Computer Science, Northwestern Polytechnical University, Xi’an, China
- 2Faculty of Data Science, City University of Macau, Macau, China
- 3State Key Laboratory of Multimodal Artificial Intelligence Systems, Institute of Automation, CAS, Beijing, China
Abstract
Recent advances in spatially resolved transcriptomics have enabled large-scale measurement of gene expression while preserving spatial context, facilitating the investigation of spatial heterogeneity within tissues. In this study, we propose SpatialGEO, a geometric-aware deep learning framework that integrates gene expression profiles with spatial coordinates to generate biologically meaningful low-dimensional embeddings, enabling the dissection of complex tissue architectures. We systematically evaluate SpatialGEO across multiple tissue types and diverse SRT platforms. Results show that SpatialGEO achieves superior performance in tissue structure dissection and data denoising compared to state-of-the-art methods. Moreover, when applied to human breast cancer samples, SpatialGEO precisely delineates the tumor microenvironment and uncovers molecular heterogeneity within tumors and intercellular communication between invasive ductal carcinoma and tumor edge. In mouse embryogenesis, SpatialGEO accurately reconstructs spatiotemporal tissue architectures, highlighting organ-specific developmental programs and elucidating molecular drivers of early neural development.
Related
- Comments on "SpatialDG: a novel spatial domain identification method for spatially resolved transcriptomics data based on dual-graph neural network".
- Si-scRNA-seq: Enabling ambient-stable single-cell transcriptomics through spatially confined biosilicification
- GeoEPred: A multimodal structure-aware geometric deep learning framework for Gram-negative bacterial secreted effector prediction with sequence semantics
- Deciphering the regulatory code of RNA inosine through enzymatic precision mapping and an explainable deep learning model
- DiST: disentangling structure via self-supervised tensors for heterophilous spatial transcriptomics.
BibTeX
@article{li2026geometricaware,
title = {{Geometric-aware deep learning for deciphering tissue structure from spatially resolved transcriptomics}},
author = {Xingyi Li and Xiangting Jia and Dongmin Zhao and Jialuo Xu and Gaoyuan Du and Yang Qi and Yingfu Wu and Yiqi Chen and Junnan Zhu and Jia Gu and Xuequn Shang},
journal = {Communications Biology},
year = {2026},
month = jul,
doi = {10.1038/s42003-026-10667-1},
url = {https://doi.org/10.1038/s42003-026-10667-1},
note = {CC BY-NC-ND 4.0},
}
RIS
TY - JOUR
TI - Geometric-aware deep learning for deciphering tissue structure from spatially resolved transcriptomics
AU - Xingyi Li
AU - Xiangting Jia
AU - Dongmin Zhao
AU - Jialuo Xu
AU - Gaoyuan Du
AU - Yang Qi
AU - Yingfu Wu
AU - Yiqi Chen
AU - Junnan Zhu
AU - Jia Gu
AU - Xuequn Shang
PY - 2026
DA - 2026/07/11
JO - Communications Biology
DO - 10.1038/s42003-026-10667-1
UR - https://doi.org/10.1038/s42003-026-10667-1
AB - Recent advances in spatially resolved transcriptomics have enabled large-scale measurement of gene expression while preserving spatial context, facilitating the investigation of spatial heterogeneity within tissues. In this study, we propose SpatialGEO, a geometric-aware deep learning framework that integrates gene expression profiles with spatial coordinates to generate biologically meaningful low-dimensional embeddings, enabling the dissection of complex tissue architectures. We systematically evaluate SpatialGEO across multiple tissue types and diverse SRT platforms. Results show that SpatialGEO achieves superior performance in tissue structure dissection and data denoising compared to state-of-the-art methods. Moreover, when applied to human breast cancer samples, SpatialGEO precisely delineates the tumor microenvironment and uncovers molecular heterogeneity within tumors and intercellular communication between invasive ductal carcinoma and tumor edge. In mouse embryogenesis, SpatialGEO accurately reconstructs spatiotemporal tissue architectures, highlighting organ-specific developmental programs and elucidating molecular drivers of early neural development.
SN - 2399-3642
ER -