Fig. 1

Schematic diagram of SOAPy. a Data Preprocessing module that imports data, generates cell network, and identifies spatial domains. Data from different spatial omics technologies are converted to a unified data structure. Cell network could be built by any of the four methods. Spatial domains are inferred by unsupervised learning from expression and morphological data, or supervised classification based on the expression of signature genes. b Molecular Spatial Dynamics module. Spatial tendency analysis finds genes or cells whose expression changes with spatial distance to the given region. Spatiotemporal Pattern analysis performs a tensor decomposition to discover the major modes of variation in space and time. c Spatial Architecture module. Neighborhood and infiltration analysis find spatial proximal cell types. Spatial composition reveals conserved C-niches to delineate the cell type composition of the neighbors. d Spatial Communication module that combines spatial distance, expression level, and action mechanism of ligand-receptors (LRs) to infer cell interactions. The contact and secreted LRs are considered for short-range and long-range cell communications, respectively. Results at cell/spot level indicate the heterogeneous interaction among different spatial locations; they are further integrated to cell type-level to report significant LRs for any two cell types