SubPoE: a subgraph-induced Product-of-Experts framework for spatial multi-omics integration
5 affiliations
- 1School of Medicine and Health, Harbin Institute of Technology, Harbin, China
- 2Zhengzhou Advanced Research Institute of Harbin Institute of Technology, Zhengzhou, China
- 3School of Computer Science and Technology, Zhoukou Normal University, Zhoukou, China
- 4School of Artificial Intelligence, Henan University, Zhengzhou, China
- 5College of Animal Science and Technology, Henan University of Animal Husbandry and Economy, Zhengzhou, China
Abstract
Introduction: Spatial multi‐omics technologies jointly profile multiple molecular layers in situ, enabling integrated characterization of tissue organization and cellular states. However, many graph-based approaches remain computationally demanding because they rely on repeated message passing over global graphs, whereas naive local sampling may disrupt cell-level correspondence when multimodal features and graph topologies are not induced synchronously. Moreover, simple concatenation or fixed-weight fusion does not explicitly account for unequal uncertainty between expression‐derived and spatial evidence, making the learned representations sensitive to feature corruption, technical noise and modality inconsistency.
Methods: We present SubPoE, a subgraph-induced Product-of-Experts framework for spatial multi‐omics integration. SubPoE synchronously induces aligned local subgraphs from modality‐specific molecular similarity graphs and a shared spatial graph, preserving molecular‐similarity topology correspondence while reducing per‐iteration graph complexity. Dual‐topology variational graph encoders model expression-derived structure and spatial context as probabilistic experts, which are adaptively fused according to their uncertainty through a Product‐of‐Experts formulation. A composite self-supervised objective further improves robustness and cross‐modal alignment.
Results: Across eight public paired spatial multi‐omics datasets, SubPoE achieves competitive and comparatively stable performance against existing methods. Ablation, sensitivity, robustness and efficiency experiments further support the contributions of synchronous subgraph induction and uncertainty‐aware fusion.
Discussion: These results establish SubPoE as an effective approach for learning topology‐preserving, uncertainty-aware representations of heterogeneous spatial multi‐omics data.
Related
- Efficient structural information integration into protein language models with implicit mixture-of-experts ProBEiT
- Multi-omics survival prediction under modality incompleteness via dynamic modality graphs and product-of-experts fusion.
- Regulation of immune signal integration and memory by inflammation-induced chromosome conformation
- AI-guided design and biomaterial integration for next-generation extracellular vesicle therapeutics: an evidence-weighted translational framework.
- Connecting high‐throughput biodiversity inventories: Opportunities for a site‐based genomic framework for global integration and synthesis
BibTeX
@article{li2026subpoe,
title = {{SubPoE: a subgraph-induced Product-of-Experts framework for spatial multi-omics integration}},
author = {Yuanfang Li and Wentao Gao and Junhao Ren and Qianran Wei and Xinlei Wang and Liheng Zhang and Ruiting Li and Yadong Wang and Xiao Zhang},
journal = {Frontiers in Genetics},
year = {2026},
month = oct,
doi = {10.3389/fgene.2026.1977260},
url = {https://doi.org/10.3389/fgene.2026.1977260},
note = {CC BY 4.0},
}
RIS
TY - JOUR
TI - SubPoE: a subgraph-induced Product-of-Experts framework for spatial multi-omics integration
AU - Yuanfang Li
AU - Wentao Gao
AU - Junhao Ren
AU - Qianran Wei
AU - Xinlei Wang
AU - Liheng Zhang
AU - Ruiting Li
AU - Yadong Wang
AU - Xiao Zhang
PY - 2026
DA - 2026/10/22
JO - Frontiers in Genetics
DO - 10.3389/fgene.2026.1977260
UR - https://doi.org/10.3389/fgene.2026.1977260
AB - Introduction: Spatial multi‐omics technologies jointly profile multiple molecular layers in situ, enabling integrated characterization of tissue organization and cellular states. However, many graph-based approaches remain computationally demanding because they rely on repeated message passing over global graphs, whereas naive local sampling may disrupt cell-level correspondence when multimodal features and graph topologies are not induced synchronously. Moreover, simple concatenation or fixed-weight fusion does not explicitly account for unequal uncertainty between expression‐derived and spatial evidence, making the learned representations sensitive to feature corruption, technical noise and modality inconsistency. Methods: We present SubPoE, a subgraph-induced Product-of-Experts framework for spatial multi‐omics integration. SubPoE synchronously induces aligned local subgraphs from modality‐specific molecular similarity graphs and a shared spatial graph, preserving molecular‐similarity topology correspondence while reducing per‐iteration graph complexity. Dual‐topology variational graph encoders model expression-derived structure and spatial context as probabilistic experts, which are adaptively fused according to their uncertainty through a Product‐of‐Experts formulation. A composite self-supervised objective further improves robustness and cross‐modal alignment. Results: Across eight public paired spatial multi‐omics datasets, SubPoE achieves competitive and comparatively stable performance against existing methods. Ablation, sensitivity, robustness and efficiency experiments further support the contributions of synchronous subgraph induction and uncertainty‐aware fusion. Discussion: These results establish SubPoE as an effective approach for learning topology‐preserving, uncertainty-aware representations of heterogeneous spatial multi‐omics data.
SN - 1664-8021
ER -