Skip to contents

Export a Seurat, SingleCellExperiment or bulk matrix to .h5ad

Source: R/quasar_importer_exporter.R
quasar_h5adexporter.Rd

Single-cell objects contribute a features x cells assay; bulk data is a plain samples x genes matrix (n_obs x n_vars, the anndata convention). An optional cell-type fractions matrix (samples x cell types) can be written alongside — useful for deconvolution ground truth.

Usage

quasar_h5adexporter(
  object,
  filename,
  assay = NULL,
  layer = "counts",
  obs = NULL,
  var = NULL,
  fractions = NULL,
  fractions_to = c("obsm", "obs", "both"),
  obsm = NULL,
  bulk_orientation = c("samples_x_genes", "genes_x_samples"),
  export_reductions = TRUE,
  overwrite = TRUE,
  verbose = TRUE
)

Arguments

object

a Seurat, SingleCellExperiment, or a matrix / Matrix / data.frame (treated as bulk).

filename

output path (overwritten unless overwrite=FALSE).

assay

assay to export. Seurat: assay name (default = active assay). SCE: assay name (default "counts" if present, else the first). Ignored for bulk.

layer

Seurat layer/slot to pull the matrix from (default "counts").

obs, var

optional per-observation / per-variable metadata (data.frame or matrix). Mainly for bulk, where the object carries no metadata; row counts must match the matrix.

fractions

optional cell-type fraction matrix, samples x cell types (rows = observations). Column names become cell-type labels.

fractions_to

where to place the fractions: "obsm" (as adata.obsm['fractions']), "obs" (spread across obs columns), or "both".

obsm

optional named list of extra per-observation matrices (bulk only; for single-cell use export_reductions).

bulk_orientation

orientation of a bulk matrix: "samples_x_genes" (default) or "genes_x_samples".

export_reductions

logical, export Seurat reductions / SCE reducedDims into adata.obsm.

overwrite

logical, overwrite an existing file.

verbose

logical, print a start line and a summary block.

Value

(invisibly) the output filename.

Examples

if (FALSE) { # \dontrun{
## --- tiny bulk export (6 samples x 20 genes) + ground-truth fractions ----
set.seed(1)
bulk <- matrix(rpois(6 * 20, 5), nrow = 6, ncol = 20,
               dimnames = list(paste0("sample_", 1:6),
                               paste0("ENSG", 1:20)))
frac <- matrix(runif(6 * 3), nrow = 6,
               dimnames = list(rownames(bulk),
                               c("Tcell", "Bcell", "Mono")))
frac <- frac / rowSums(frac)                         # rows sum to 1
quasar_h5adexporter(bulk, tempfile(fileext = ".h5ad"),
                    fractions = frac, fractions_to = "both")

## --- tiny SingleCellExperiment export (15 genes x 8 cells) ---------------
sce <- SingleCellExperiment::SingleCellExperiment(
  assays = list(counts = matrix(rpois(15 * 8, 2), nrow = 15,
                dimnames = list(paste0("gene", 1:15),
                                paste0("cell", 1:8)))))
quasar_h5adexporter(sce, tempfile(fileext = ".h5ad"))
} # }