Skip to contents

DO.scVI

DO.scVI.Rd

This function will run the scVI Integration from the scVI python package. It includes all parameters from the actual python package and runs it by using an internal python script. The usage of a gpu is incorporated and highly recommended.

Usage

DO.scVI(
  Seu_object,
  batch_key,
  layer_counts = "counts",
  layer_logcounts = "logcounts",
  categorical_covariates = NULL,
  continuos_covariates = NULL,
  n_hidden = 128,
  n_latent = 30,
  n_layers = 3,
  dispersion = "gene-batch",
  gene_likelihood = "zinb",
  get_model = FALSE,
  ...
)

Arguments

Seu_object

Seurat object with annotation in meta.data

...

additional arguments for scvi.model.SCVI.

batch_key:

meta data column with batch information.

layer_counts:

layer with counts. Raw counts are required.

layer_logcounts:

layer with log-counts. Log-counts required for calculation of HVG.

categorical_covariates:

meta data column names with categorical covariates for scVI inference.

continuos_covariates:

meta data column names with continuous covariates for scVI inference.

n_hidden:

number of hidden layers.

n_latent:

dimensions of the latent space.

n_layers:

number of layers.

dispersion:

dispersion mode for scVI.

gene_likelihood:

gene likelihood.

get_model:

return the trained model.

Value

Seurat Object with dimensionality reduction from scVI

Examples

if (FALSE) { # \dontrun{
# Run scVI using the 'orig.ident' column as the batch key
Seu_object <- DO.scVI(Seu_object, batch_key = "orig.ident")
} # }