SpatialPCA is a spatially aware dimension reduction method that aims to infer a low dimensional representation of the gene expression data in spatial transcriptomics. SpatialPCA builds upon the probabilistic version of PCA, incorporates localization information as additional input, and uses a kernel matrix to explicitly model the spatial correlation structure across tissue locations.
Simulation
DLPFC dataset
Slide-Seq mouse cerebellum dataset
Slide-Seq V2 mouse hippocampus dataset
Human breast cancer ST dataset
Other source data could be downloaded from here.