Skip to contents

Pre-scan multiple 10x CellRanger h5 files for multi-sample loading

prescan_tenx_h5_files.Rd

Walks each input file, reads the per-file feature ids, restricts to the chosen feature_type (V3 only; ignored for V2), builds the intersection or union universe of gene ids, and returns the file tasks expected by load_multi_tenx_h5().

V2 and V3 files can be mixed in a single batch. Gene matching is done on the feature id (ensembl-style) since gene symbols collide.

Usage

prescan_tenx_h5_files(
  h5_paths,
  feature_type = "Gene Expression",
  gene_universe = c("intersection", "union"),
  .verbose = TRUE
)

Arguments

h5_paths

Character vector of file paths to 10x h5 files. If names are provided, these will be used as experimental identifiers.

feature_type

String. Modality to keep across all files (V3 only; ignored for V2). Defaults to "Gene Expression".

gene_universe

One of "intersection" or "union".

.verbose

Boolean. Controls verbosity.

Value

A list with:

  • universe - Character vector of gene ids in the universe.

  • universe_size - Length of the universe.

  • file_tasks - Named list of per-file task structures, each containing exp_id, h5_path, version, no_cells, no_genes, feature_type and gene_local_to_universe (integer vector, NA for features outside the universe / non-target modality, 0-indexed).

Examples

# gene universe across two 10x h5 files
data <- generate_single_cell_test_data(
  syn_data_params = params_sc_synthetic_data(n_cells = 200L, n_genes = 40L)
)
features <- data.table::data.table(
  id = data$var$gene_id,
  name = data$var$ensembl_id,
  feature_type = "Gene Expression"
)
files <- c(a = tempfile(fileext = ".h5"), b = tempfile(fileext = ".h5"))
for (f in files) {
  write_tenx_h5_sc(f, data$counts, data$obs$cell_id, features)
}
scan_res <- prescan_tenx_h5_files(h5_paths = files, .verbose = FALSE)
scan_res$universe_size
#> [1] 40

unlink(files)