GASTON - Mapping the topography of spatial gene expression with interpretable deep learning

GASTON-Mix is a spatial mixture-of-experts (MoE) model for learning domain-specific _topographic maps_ of a tissue slice from spatially resolved transcriptomics (SRT) data.

GASTON-Mix model architecture

  • Learn spatial domains in tissue slice, i.e. tissue geometry

  • Learn 1-d coordinate that varies smoothly across each domain, providing local topographic map of gene expression in the domain.

  • Modeling continuous gradients of gene expression for individual genes, e.g. gradients of metabolism in cancer

Manuscript

Please see our manuscript for more details.

Getting started with GASTON-Mix

  • Browse Tutorial for a quick start guide to GASTON.

  • Discuss usage and issues on github.