A deep learning framework, CroSP, based on a multi-view graph autoencoder that integrates gene expression profiles, spatial locations, and tissue morphology information into a low-dimensional latent representation that supports the use of CroSP for spatial biological analysis in healthy and diseased tissues.
Abstract
MOTIVATION
The rapid advancement of spatial transcriptomics (ST) has established a foundation for the complementary integration of multimodal information from gene expression profiles, spatial localization, and histological images. However, noise and data heterogeneity complicate the characterization of spatial relationships and tissue structure.
RESULTS
In this study, we develop a deep learning framework, CroSP, based on a multi-view graph autoencoder. The framework integrates gene expression profiles, spatial locations, and tissue morphology information into a low-dimensional latent representation. Cross-view contrastive learning and metric learning coordinate the complementary views. Across datasets from multiple platforms and spatial resolutions, CroSP improved spatial domain identification and delineated brain and olfactory bulb structures. Analyses using established downstream methods further characterized spatially variable genes and the spatial heterogeneity, immune microenvironments, and cell-cell communication in human breast cancer tissue. These results support the use of CroSP for spatial biological analysis in healthy and diseased tissues.
AVAILABILITY AND IMPLEMENTATION
Source code and tutorials: https://github.com/Biowust/CroSP.git. Archived code: https://doi.org/10.5281/zenodo.22675113. Public raw datasets: https://doi.org/10.5281/zenodo.22533825.
This work introduces SpaFactor, a lightweight and efficient low-rank morphology-program-gene factorization framework that demonstrates that lightweight joint modeling of tissue context and gene programs can improve both predictive accuracy and biological fidelity.
Shi-Ting Ruan, Xi-Tong Ling, Qi-Ming He et al.· arXiv.org· 0 citations
Spatial omics enable the in situ characterization of tissue architecture by measuring molecular modalities. However, integrating these datasets remains challenging due to differences in spatial resolution, cross-modal alignment and molecular coverage. Here we show that Cross-scale spatial Omics Registration and Analysi...
Si-Yu He, Matthew Bieniosek, Jing-Tian Zhou et al.· Nature Genetics· 0 citations
High-resolution spatial transcriptomics (ST) enables subcellular spatially resolved transcript profiling but remains difficult to annotate in complex tissues because of gene coverage and super sparseness in expression. These challenges are amplified in cancer research, where neoplastic cell states and rare immune cells...
Jing-Xuan Zhang, Lei Li, Li-Meng Liu et al.· bioRxiv· 0 citations
Spatial transcriptomics enables high-throughput gene expression profiling while preserving spatial information, offering valuable insights into tissue architecture and cellular organization. Methods that achieve single-cell or subcellular spatial resolution typically rely on predefined gene panels, which limits genome-...
A. Abir, Muhtasim Noor Alif, Wei Zhang· bioRxiv· 0 citations
We use cookies to run the site and, with your consent, for analytics and to show ads.
See our Cookie Policy.