Identification of perturbation-responsive regions and genes in comparative spatial transcriptomics atlases

We introduce Vespucci, a machine-learning method to identify perturbation-responsive regions, genes and gene programs within comparative spatial transcriptomics atlases. We validate Vespucci on simulated and published datasets and show that it outperforms 19 published computational methods for spatial transcriptomics. We apply Vespucci to expose the spatial organization of gene programs activated by therapies that guide repair of the injured spinal cord.