Github.com/uhlerlab/spatialfusion/unimodal-embeddings-for-spatialfusion

Automatic Tools and Workflows
1

Unimodal Embeddings for SpatialFusion

This workflow generates the unimodal embedding inputs used by SpatialFusion from:

  • spatial transcriptomics data with scGPT
  • H&E / whole-slide imaging data with UNI

By default, the workflow runs both embedding modes in parallel and returns:

  • scGPT.parquet
  • UNI.parquet

Main inputs

  • adata: AnnData (.h5ad) file with spatial coordinates stored in adata.obsm["spatial"]
  • input_is_log_normalized: whether the AnnData expression values
This is a companion discussion topic for the original entry at github.com/uhlerlab/spatialfusion/unimodal-embeddings-for-spatialfusion