Deciphering spatial domains from spatial multi-omics with SpatialGlue

Integration of multiple data modalities in a spatially informed manner remains an unmet need for exploiting spatial multi-omics data. Here, we introduce SpatialGlue, a novel graph neural network with dual-attention mechanism, to decipher spatial domains by intra-omics integration of spatial location and omics measurement followed by cross-omics integration. We demonstrate that SpatialGlue can more accurately resolve spatial domains at a higher resolution across different tissue types and technology platforms, to enable biological insights into cross-modality spatial correlations. 4 samples : 2 In-house human lymph node and 2 In-house tonsil datasets generated using 10x Genomics Visium RNA and protein co-profiling technology (section A1 and D1). Lymph node section A1 consists of 3,484 spots, 18,085 genes and 31 proteins; Lymph node section D1 consists of 3,359 spots, 18,085 genes and 31 proteins; Tonsil section A1 consists of 4337 spots, 18085 genes and 31 proteins; Tonsil section D1 consists of 4,519 spots, 18,085 genes and 31 proteins. Please note that processed data generated from both transcriptomics and protein profiling is linked to the corresponding transcriptomics sample records.