Deciphering spatial domains from spatial multi-omics with SpatialGlue

Abstract

Advances in spatial omics technologies now allow multiple types of data to be acquired from the same tissue slice. To realize the full potential of such data, we need spatially informed methods for data integration. Here, we introduce SpatialGlue, a graph neural network model with a dual-attention mechanism that deciphers spatial domains by intra-omics integration of spatial location and omics measurement followed by cross-omics integration. We demonstrated SpatialGlue on data acquired from different tissue types using different technologies, including spatial epigenome–transcriptome and transcriptome–proteome modalities. Compared to other methods, SpatialGlue captured more anatomical details and more accurately resolved spatial domains such as the cortex layers of the brain. Our method also identified cell types like spleen macrophage subsets located at three different zones that were not available in the original data annotations. SpatialGlue scales well with data size and can be used to integrate three modalities. Our spatial multi-omics analysis tool combines the information from complementary omics modalities to obtain a holistic view of cellular and tissue properties.

Biophysical Simulation Enables Multi-Scale Segmentation and Atlas Mapping for Top-Down Spatial Omics of the Nervous System

“…These 'bottom-up' approaches often employ clustering algorithms which learn locally and globally variable gene expression patterns. Transcript density is used to either identify distinct spatial domains with discontinuous changes in gene expression or assume continuous expression and fit a function to model the expression of a transcript across space [33][34][35][36] . GASTON models the transcriptomic space as a topographic map, fitting the spatial landscape based on expression density…”

“…. Alternatively, bottom-up approaches may also resolve spatial domains via the integration of multi-omics modalities, such as SpatialGlue and MIP-Seq 34,39 , which require access to various omics datasets for a given slice. There are even techniques to integrate two different species with gene expression information 40 .…”

Predicting lncRNA–Disease Associations Based on a Dual-Path Feature Extraction Network with Multiple Sources of Information Integration

“…In recent years, deep learning models, due to their powerful representation learning ability, have greatly promoted the development of bioinformatics and are increasingly being used in various bioinformatics analysis tasks. For example, SpatialGlue 21 applied graph learning and attention mechanisms to the process of spatial transcriptomics data integration, unleashing the full potential of multimodal data. DrlGCL 22 used graph learning and contrastive learning methods to infer potential associations between drugs and diseases, not only improving the accuracy of association prediction, but also saving time and costs.…”