AddaGCN: Spatial transcriptomics deconvolution using graph convolutional networks with adversarial discriminative domain adaptation

Abstract

The rapid advancement of spatial transcriptomics has substantially improved our understanding of the spatial architecture and gene expression heterogeneity within tissues. However, many spatial transcriptomics techniques can not reach single-cell resolution, instead measuring gene expression profiles from mixtures of potentially heterogeneous cell types. Here we propose AddaGCN, a robust deconvolution method to infer cell type composition from spatial transcriptomic data. AddaGCN leverages graph convolutional networks to incorporate spatial information and adopts an adversarial discriminative domain adaptation approach to mitigate batch effects between spatial and single-cell reference data. Comprehensive analyses of real data generated by diverse technology platforms demonstrate AddaGCN’s superior performance and robustness in cell-type deconvolution compared to other methods. These analyses further reveal AddaGCN’s potential to uncover spatiotemporal changes during tissue development and to characterize the tumor microenvironment.