# GATCL: graph attention network meets contrastive learning for spatial domain identification

## TL;DR

GATCL is a new deep learning method that improves the identification of spatial domains in tissues by combining graph attention networks and contrastive learning.

## Contribution

GATCL introduces a novel integration of graph attention networks and contrastive learning for more accurate spatial domain identification.

## Key findings

- GATCL outperforms seven existing methods across six datasets and six evaluation metrics.
- The graph attention mechanism allows dynamic weighting of neighboring spots for better modeling of cellular architecture.
- Cross-modal contrastive learning enhances alignment between different omics data types.

## Abstract

Spatial domain identification is an essential task for revealing spatial heterogeneity within tissues, providing insights into disease mechanisms, tissue development, and the cellular microenvironment. In recent years, spatial multi-omics has emerged as the new frontier in spatial domain identification that offers deeper insights into the complex interplay and functional dynamics of heterogeneous cell communities within their native tissue context. Most existing methods rely on static graph structures that treat all neighboring cells uniformly, failing to capture the nuanced cellular interactions within the microenvironment and thus blurring functional boundaries. Furthermore, cross-modal reconstruction performance is often degraded by overfitting to modality-specific noise, which may impair the precise delineation of spatial domains. Therefore, we present GATCL, a novel deep learning framework that integrates a graph attention network with contrastive learning (CL) for robust spatial domain identification. First, GATCL leverages the graph attention mechanism to dynamically assign weights to neighboring spots, adaptively modeling the complex cellular architecture. Second, it implements a cross-modal CL strategy that forces representations from the same spatial location to be similar while pushing those from different locations apart, thereby achieving robust alignment between modalities. Comprehensive experiments across six distinct datasets (spanning transcriptome, proteome, and chromatin) reveal that GATCL is superior to seven representative methods across six key evaluation metrics.