SCG: Spatially Co-Expressed Gene Identification through Spatially Varying Networks

Abstract

Spatial transcriptomics has enabled the advancement of gene expression analysis, yet spatial co-expression remains understudied. We introduce spatial covariance regression (SCR), a scalable Bayesian factor-model-based framework for estimation of spatially-resolved gene co-expression networks across tissue domains. These networks provide the spatial map of gene-gene correlations and enable the identification of spatially co-expressed genes (SCGs), which serve as potential prognostic biomarkers and therapeutic targets.