Sep 2026· Research Square· 0 citations· 27 references
Medicine
TL;DR
This work introduces spaGFM, a graph foundation model that serializes cellular neighborhoods through random walks to generate transformer-compatible representations of tissue organization that transfer across tissues, disease contexts, and spatial technologies.
Abstract
Foundation models offer a promising paradigm for modeling spatial transcriptomics, but capturing tissue context over cellular graphs makes training at scale challenging. We introduce spaGFM, a graph foundation model that serializes cellular neighborhoods through random walks to generate transformer-compatible representations of tissue organization. By replacing classical graph message passing with augmented self-supervised neighborhood reconstruction, spaGFM captures higher-order spatial context while allowing scaling at the atlas level. Pretrained on 43.9 million cells from 132 image-based spatial transcriptomics datasets, spaGFM produces robust representations at both the neighborhood and cell level that transfer across tissues, disease contexts, and spatial technologies. SpaGFM identifies tertiary lymphoid structures across cancer cohorts and spatial platforms, captures cell-level transcriptomic perturbation responses associated with T cell vicinity in spatial CRISPR experiments, and characterizes glomerular organization associated with pathological grade in diabetic kidney disease biopsies. Overall, spaGFM establishes a graph foundation-model framework for learning cellular organization and its functional consequences.
Spatial transcriptomics captures molecular states within cells and their organisation in tissue. However, integrating fine-grained gene information with spatial context at scale remains challenging for existing foundation models. Here we present NexuST, a hierarchical foundation model that repeatedly interleaves gene-l...
Hai-Ping Liu, Qian Zhao, Li-Jing Lin et al.· bioRxiv· 0 citations
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
MultiSpaceNet is presented, a graph-based framework for joint representation learning from paired spatial transcriptomic and proteomic data that outperforms nine published state-of-the-art methods in spatial domain identification and preserves biological structure across replicate sections better than all compared alte...
Zheng-Qiang Zhang, Bing-Hong Chen, Jing-Yi Bai et al.· Bioinformatics· 0 citations
This work introduces a geometric framework for analyzing the spatiotemporal evolution of gene expression networks through embeddings in Gromov--Wasserstein (GW) space and demonstrates that hypernetwork representations successfully record salient biological changes across time.
M. Oliver, K. Hohmeier, Tuyến Trần et al.· 0 citations
Accurately identifying spatial domains based on spatial transcriptomics (ST) data can greatly promote our understanding of cellular composition and tissue organization. While graph neural networks (GNNs) have shown significant advancements in spatial clustering, they tend to be insensitive to noisy edges, leading to in...
Li-Hong Peng, Long Yang, Min Chen et al.· Genomics, Proteomics & Bioin...· 0 citations
Spatial transcriptomics now profiles patient cohorts at single-cell resolution, enabling analysis of disease-associated cell organization in situ. However, discovering such spatial biomarkers remains challenging because relevant structures occur at unknown scales and cell-or niche-level annotations are rarely available...
Hoeyoung Kim, Donghee Kim, Sangwook Jung et al.· bioRxiv· 0 citations
We use cookies to run the site and, with your consent, for analytics and to show ads.
See our Cookie Policy.