Skip to contents

Load in h5ad to SingleCells

load_h5ad.Rd

This function takes an h5ad file and loads the obs and var data into the DuckDB of the SingleCells class and the counts into a Rust-binarised format for rapid access. During the reading in of the counts, the log CPM transformation will occur automatically.

Usage

load_h5ad(
  object,
  h5_path,
  sc_qc_param = params_sc_min_quality(),
  streaming = 1L,
  raw_count_slot = c("auto", "X", "raw.X", "layers.counts"),
  cell_id_col = NULL,
  batch_size = 1000L,
  max_genes_in_memory = 2000L,
  cell_batch_size = 100000L,
  .verbose = TRUE
)

Arguments

object

SingleCells class.

h5_path

File path to the h5ad object you wish to load in.

sc_qc_param

List. Output of params_sc_min_quality(). A list with the following elements:

  • min_unique_genes - Integer. Minimum number of genes to be detected in the cell to be included.

  • min_lib_size - Integer. Minimum library size in the cell to be included.

  • min_cells - Integer. Minimum number of cells a gene needs to be detected to be included.

  • target_size - Float. Target size to normalise to. Defaults to 1e5.

streaming

Integer. 0L -> all cells loaded in memory then transposed (fastest, highest memory), 1L -> light streaming with cell batching, 2L -> heavy streaming with memory upper boundaries on the gene side. Controls memory pressure during the CSR-to-CSC conversion. Defaults to 1L.

raw_count_slot

Where raw counts live. "auto" detects per file via detect_raw_count_slot(); otherwise one of "X", "raw.X", "layers.counts".

cell_id_col

Optional string. If a specific column in the h5ad obs data is representing the cell identifiers, you can specify it here.

batch_size

Integer. Cell batch size when streaming = 1L. Defaults to 1000L.

max_genes_in_memory

Integer. Maximum genes held in memory at once when streaming = 2L. Defaults to 2000L.

cell_batch_size

Integer. Cell batch size when streaming = 2L. Defaults to 100000L.

.verbose

Boolean. Controls the verbosity of the function.

Value

It will populate the files on disk and return the class with updated shape information.

Examples

# round trip through a sparse h5ad file
data <- generate_single_cell_test_data(
  syn_data_params = params_sc_synthetic_data(n_cells = 200L, n_genes = 40L)
)
f_path <- tempfile(fileext = ".h5ad")
write_h5ad_sc(f_path, data$counts, data$obs, data$var, .verbose = FALSE)
dir_data <- tempfile("sc_h5ad")
dir.create(dir_data, recursive = TRUE)
sc <- load_h5ad(
  object = SingleCells(dir_data = dir_data),
  h5_path = f_path,
  sc_qc_param = params_sc_min_quality(
    min_unique_genes = 5L,
    min_lib_size = 25L,
    min_cells = 5L
  ),
  streaming = 0L,
  .verbose = FALSE
)
dim(sc)
#> [1] 200  40

unlink(c(f_path, dir_data), recursive = TRUE, force = TRUE)