# AugGCL: Multimodal graph learning for spatial transcriptomics analysis with enhanced gene and morphological data

## TL;DR

AugGCL is a new method for spatial transcriptomics that improves the accuracy of identifying tissue structures by combining gene expression and morphological data.

## Contribution

AugGCL introduces a multimodal graph learning framework that integrates gene and image data to enhance spatial domain identification.

## Key findings

- AugGCL outperforms baseline methods on multiple datasets, including human brain, breast cancer, and mouse embryo.
- The method improves spatial domain recognition accuracy by integrating gene expression with tissue morphology.
- Downstream analyses validated the method's effectiveness in cell annotation and functional enrichment.

## Abstract

Spatial transcriptomics enables the measurement of gene expression in intact tissues. Despite this, reconstructing anatomically accurate spatial domains remains challenging, primarily due to expression sparsity, complex tissue architecture that is characterized by sharp boundaries and long-range continuity, and weak spatial signals. Traditional pipelines typically rely on expression-driven clustering and spatial smoothing, which underperform at boundaries and in sparse regions while neglecting morphological information. To address these challenges, AugGCL is proposed, an augmented graph-convolutional learning framework that enhances spatial structure decoding and gene expression reconstruction through targeted augmentation of both gene and image data. A key component of AugGCL is neighborhood information aggregation mechanism, which integrates expression similarity and spatial proximity to construct a weighted graph and an enhanced expression matrix, addressing sparsity without sacrificing boundary clarity. Additionally, a two stream weighted graph convolutional network jointly models refined gene features and image-derived morphological information, with image-aware auxiliary reconstructions enhancing weak spatial signals and sharpening boundaries. On datasets from the human dorsolateral prefrontal cortex, breast cancer, and mouse embryo, AugGCL outperforms baseline methods across multiple metrics, showing robustness and generalization across a range of datasets. Downstream analysis validated the reliability of the method, confirming its effectiveness in cell annotation, functional enrichment, and mechanistic studies. AugGCL generates clearer spatial domains and significantly advances the application of spatial transcriptomics in tissue structure and disease research.

Spatial transcriptomics is an important technique for revealing tissue structure and disease mechanisms. However, existing spatial domain identification methods have not fully exploited the spatial information embedded in the data, especially when it comes to detailed exploration of tissue structure. This study presents a new tool for spatial domain identification, which makes full use of various types of spatial transcriptomic data from multiple perspectives to accurately identify real cell groupings. The method significantly improves spatial domain recognition accuracy by integrating gene expression with tissue morphology analysis. Experimental results show that this tool not only accurately identifies spatial domains but also provides strong support for the comprehensive exploration of biological tissue structure. On this basis, a series of downstream analyses, including volcano plot generation, functional enrichment analysis, and gene heatmap visualization, were performed. These analyses not only validated the effectiveness of the method but also revealed the functional characteristics and expression patterns of cells within spatial domains. This step further confirmed the broad application potential of the method in cell type annotation, functional enrichment, and mechanistic research, highlighting its significant potential in advancing biological and disease research.