FuseMap

Spatial integration and mapping with universal gene, cell, and tissue embeddings.

FuseMap is a deep-learning framework for spatial transcriptomics that (1) bridges single-cell or single-spot gene expression within spatial contexts and (2) consolidates various gene panels across technologies, organs, and species.

---

Contents

• Overview
• Installation
  • Installing the package
  • Downloading the pretrained models
• Tutorials
  • 1. Spatially integrate imaging-based data
  • 2. Spatially integrate imaging-based and sequencing-based data
  • 3. Spatially impute transcriptome-wide genes
  • 4. Map new datasets to existing pre-trained FuseMap model (customized)
  • 5. Map new datasets to the existing pre-trained FuseMap model (molCCF)
  • 6. Infer cell-cell communication/interactions
• About
  • Support
  • Authors
  • How to Cite

---

Quick start

Spatial integration

fusemap.spatial_integrate provides tools to integrate spatial transcriptomics data. Input data can be from any spatial transcriptomics technology, such as Visium, Slide-seq, or MERFISH. Output data can be used for downstream analysis, such as clustering, cell type identification, or spatial gene expression analysis.

Spatial mapping

fusemap.spatial_map provides tools to map spatial transcriptomics data to a universal gene, cell, and tissue embedding space.

Step-by-step guide

Check out our detailed tutorials on how to use FuseMap for spatial integration and mapping.