The tools behind the papers. Everything here is open source and developed with the Lotfollahi Lab.
End-to-end analysis of spatial multi-omics data. A graph variational autoencoder that learns cell-niche representations which stay interpretable, because each latent dimension is tied to a known intercellular communication program.
A spatial transcriptomics foundation model built on a graph transformer, modelling human tissue from the single cell up to the whole section.
Discrete tokenization for spatial transcriptomics tissue sections, learning compact cell and niche codes that support in-painting of missing tissue regions.
Separates microenvironment-induced from cell-intrinsic gene expression, so that tissue microenvironments can be both mapped and reprogrammed.
Tissue reconstruction from gene expression via embedding-distance flow matching — generating spatial structure rather than only reading it.