Skip to contents

Identify HVGs without mutating object state

get_hvg_data_sc.Rd

Like find_hvg_sc() but does not mutate object. Returns a data.table with per-gene HVG statistics plus is_hvg/hvg_rank for the top hvg_no genes. Useful for computing HVGs on a subset of cells (e.g. a specific cell type) for downstream methods like NMF, without overwriting the HVGs stored on the object.

Usage

get_hvg_data_sc(
  object,
  cell_ids = NULL,
  hvg_no = 3000L,
  hvg_params = params_sc_hvg(),
  streaming = NULL,
  .verbose = TRUE
)

Arguments

object

SingleCells or MetaCells class.

cell_ids

Optional character. Cell ids (or meta cell ids) to restrict the HVG calculation to. If NULL, uses get_cells_to_keep() for SingleCells and all meta cells for MetaCells.

hvg_no

Integer. Number of top HVGs to flag. Defaults to 3000L.

hvg_params

List, see params_sc_hvg().

streaming

Optional Boolean. Stream the data. Ignored for MetaCells.

.verbose

Boolean or integer. Verbosity.

Value

data.table with gene_idx, gene_id, the HVG statistics returned by the Rust HVG function, an is_hvg boolean and an hvg_rank integer (NA for non-HVGs).

Examples

# HVG statistics without touching the object
sc <- demo_single_cells(prepped = FALSE)
dt <- get_hvg_data_sc(sc, hvg_no = 20L, .verbose = FALSE)
head(dt[(is_hvg), c("gene_id", "hvg_rank")], 3)
#>    gene_id hvg_rank
#>     <char>    <int>
#> 1: gene_04       20
#> 2: gene_05        3
#> 3: gene_07       19

unlink(sc@dir_data, recursive = TRUE, force = TRUE)