Total Variational Inference

This page is under construction. For now, please learn more about totalVI in our user guide.

# totalVI is for end-to-end analysis of CITE-seq data

import anndata

import scvi

# read data and prepare for scvi-tools

adata = anndata.read("my_data.h5ad")

scvi.data.setup_anndata(adata, protein_expression_obsm_key="proteins")

model = scvi.model.TOTALVI(adata)

model.train()

# get integrated low-dimensional representation

adata.obsm["X_totalvi"] = model.get_latent_representation()

# normalized expression

adata.layers["totalvi_norm"], adata.obsm["totalvi_pro_norm"] = model.get_normalized_expression()

# differential expression

results = model.differential_expression(

groupby="cell types",

group1="CD4",

group2="CD8"

)