Global Context-aware Representation Learning for Spatially Resolved Transcriptomics

2025-05-28

AI Summary

Existing spatially resolved transcriptomics (SRT) graph models struggle to accurately characterize spots—especially those near spatial domain boundaries—due to overreliance on local neighborhood similarity while neglecting cross-region long-range dependencies. To address this, we propose Spotscape: first, a “similarity telescope” module that learns adaptive similarity metrics to capture global contextual information; second, a similarity scaling strategy that jointly normalizes intra- and inter-slice distances, enabling robust multi-slice alignment and integration. By breaking the locality constraint inherent in conventional graph neural networks, Spotscape achieves superior performance in both single-slice spot identification and multi-slice integration. On multiple benchmark SRT datasets, it improves clustering accuracy by over 8% on average compared to state-of-the-art methods and significantly enhances discriminative representation learning for boundary-adjacent spots.

Technology Category

• Machine Learning: Graph-based Machine Learning
• Data Mining & Knowledge Management: Graph Mining, Social Network Analysis & Community
• Computer Vision: Multi-modal Vision

Application Category

• Graph Algorithms and Modeling for the Web: Representation, reconstruction, and subgraph or motif discovery in Web-related graphs
• Search and Retrieval-Augmented AI: Web query analysis, representation and understanding
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs

Abstract

Spatially Resolved Transcriptomics (SRT) is a cutting-edge technique that captures the spatial context of cells within tissues, enabling the study of complex biological networks. Recent graph-based methods leverage both gene expression and spatial information to identify relevant spatial domains. However, these approaches fall short in obtaining meaningful spot representations, especially for spots near spatial domain boundaries, as they heavily emphasize adjacent spots that have minimal feature differences from an anchor node. To address this, we propose Spotscape, a novel framework that introduces the Similarity Telescope module to capture global relationships between multiple spots. Additionally, we propose a similarity scaling strategy to regulate the distances between intra- and inter-slice spots, facilitating effective multi-slice integration. Extensive experiments demonstrate the superiority of Spotscape in various downstream tasks, including single-slice and multi-slice scenarios. Our code is available at the following link: https: //github.com/yunhak0/Spotscape.

Problem

• Improves spot representation in spatially resolved transcriptomics
• Addresses boundary spot issues in spatial domain identification
• Enhances multi-slice integration with global similarity scaling

Innovation

• Introduces Similarity Telescope for global spot relationships
• Uses similarity scaling for intra- and inter-slice spots
• Enhances representation learning for spatial domain boundaries