Skip to contents

Convert a SingleCellExperiment to a Python AnnData object

sce_to_anndata.Rd

Creates a Python anndata.AnnData object from a SingleCellExperiment::SingleCellExperiment using a scipy.sparse.csc_matrix built with sce_assay_to_scipy_csc().

Copy semantics

  • X (the main count matrix): the dgCMatrix is first converted to a scipy.sparse.csc_matrix (shape genes × cells) via sce_assay_to_scipy_csc(), then transposed to scipy.sparse.csr_matrix (shape cells × genes) by .T. The transpose shares all data buffers with the original CSC (flags.owndata = FALSE), so no values are copied.

  • obs / var metadata: a pandas DataFrame is constructed from the R data.frame; reticulate performs one copy here, which is unavoidable for columnar R data frames → row-oriented Python objects.

  • layers / obsm / varm: passed through as-is; the caller is responsible for any copy semantics of those objects.

Usage

sce_to_anndata(
  x,
  assay = "counts",
  obs = NULL,
  var = NULL,
  layers = list(),
  obsm = list(),
  varm = list()
)

Arguments

x

A SingleCellExperiment::SingleCellExperiment.

assay

A single string: the assay to place in X (default "counts").

obs

NULL or a data.frame of cell-level metadata. Defaults to as.data.frame(colData(x)) with rownames set to colnames(x).

var

NULL or a data.frame of gene-level metadata. Defaults to as.data.frame(rowData(x)) with rownames set to rownames(x).

layers

A named list of additional assay matrices to place in AnnData$layers.

obsm

A named list of matrices added to AnnData$obsm.

varm

A named list of matrices added to AnnData$varm.

Value

A Python anndata.AnnData object (reticulate, convert = FALSE).

See also

sce_assay_to_scipy_csc(), sce_to_reticulate_anndata()

Examples

if (FALSE) { # \dontrun{
sce <- SingleCellExperiment::SingleCellExperiment(
  assays = list(counts = Matrix::rsparsematrix(200, 50, 0.1, repr = "C"))
)
adata <- sce_to_anndata(sce)
cat("obs x var:", reticulate::py_str(adata$shape), "\n")
} # }