|
Abstract:
|
Spatial gene expression technologies profile both gene expression and spatial locations of cells, revealing spatial organizations of cell types and cell-cell interactions in a spatial context. Most existing methods analyze a single spatial gene expression sample, and methods to analyze multiple spatial gene expression samples are lacking. We present POLYspace, a comprehensive framework to analyze and compare multiple spatial gene expression samples. POLYspace can be generally applied to spatial data generated by different platforms and with manual or computational annotations. POLYspace employs a flexible regression framework to identify changes in spatial interactions, spatial colocalizations, cell-type compositions, and gene expression profiles associated with a sample phenotype. Using both simulated and real spatial gene expression data, we demonstrate that POLYspace captures spatial features associated with tissue regions and developmental time stages.
|
