<i> DESpace </i> : spatially variable gene detection via differential expression testing of spatial clusters

Abstract

In recent years, various methods have been proposed to discover SVGs; notably: MERINGUE (), a density-independent approach built on the spatial auto-correlation of each gene; nnSVG (), which fits nearest-neighbor Gaussian processes; SpaGCN (), that identifies spatial domains based on a graph convolutional network, and performs a Wilcoxon rank-sum test on the mean-level expression changes between domains; SPARK (), using generalized linear mixed models with overdispersed Poisson and Gaussian distributions; SPARK-X (), from the authors of SPARK, based on a scalable non-parametric framework to te