cellflow.external.CFJaxSCVI

class cellflow.external.CFJaxSCVI(adata, n_hidden=128, n_latent=10, dropout_rate=0.1, gene_likelihood='normal', **model_kwargs)[source]

CellFlow-specific JAX scVI model.

A lightweight VAE that inherits the scvi-tools training and data infrastructure but constructs its own CFJaxVAE module, avoiding coupling to upstream JaxSCVI.__init__ kwargs.

Parameters:

  • adata (AnnData) – AnnData registered via CFJaxSCVI.setup_anndata().

  • n_hidden (int) – Number of nodes per hidden layer.

  • n_latent (int) – Dimensionality of the latent space.

  • dropout_rate (float) – Dropout rate for neural networks.

  • gene_likelihood (Literal['nb', 'poisson', 'normal']) – One of 'nb', 'poisson', 'normal'.

  • **model_kwargs – Forwarded to CFJaxVAE.

Methods

get_latent_representation([adata, indices, ...])

Return the latent representation for each cell.

get_reconstructed_expression(data[, ...])

Return the reconstructed expression for each cell.

setup_anndata(adata[, layer, batch_key])

Set up AnnData for use with CFJaxSCVI.

to_device(device)

Attributes

device