spatialvi-tools

spatialvi-tools#

spatialvi-tools is a Python package for spatial transcriptomics analysis using deep learning. It provides a unified interface for various spatial methods including deconvolution, niche analysis, and cell-cell interaction inference.

Key Features#

  • Spatial VAE Models: Variational autoencoders that incorporate spatial context

  • Deconvolution Methods: Cell type proportion estimation for bulk spatial spots

  • Niche Analysis: Modeling cellular microenvironments

  • Cell-Cell Interactions: Inference of spatial cell-cell communication

  • Integration with scvi-tools: Seamless use of scVIVA, ResolVI, and DestVI

Installation#

pip install spatialvi-tools

For development:

pip install -e ".[dev]"

Quick Start#

import spatialvi

# Load data
adata = spatialvi.data.synthetic_spatial()

# Compute spatial neighbors
spatialvi.data.compute_spatial_neighbors(adata)

# Setup and train model
SpatialVAE.setup_anndata(adata, spatial_key="spatial")
model = spatialvi.model.SpatialVAE(adata)
model.train(max_epochs=100)

# Get latent representation
latent = model.get_latent_representation()

Contents#

Indices and tables#

Contents