Subset a SingleCellExperiment by Cell Type — subsetSCE • spacedeconv

This function facilitates the subsetting of a SingleCellExperiment object to a specified number of cells, with an option to maintain an even or proportional distribution across different cell types. It offers flexibility in handling scenarios where insufficient cells of a specific type are available and ensures reproducibility through the option to set a seed for random sampling.

Usage

subsetSCE(
  sce,
  cell_type_col = "celltype_major",
  scenario = "even",
  ncells = 1000,
  notEnough = "asis",
  seed = 12345
)

Arguments

sce: A SingleCellExperiment object containing cell type annotations. The function applies subsetting operations on this object to achieve a targeted composition of cell types.
cell_type_col: A character string indicating the column name within colData of the SingleCellExperiment object that contains cell type information. This column is used to identify and categorize cells by type for subsetting.
scenario: Specifies the strategy for cell selection across types: "even" for equal numbers from each cell type, or "mirror" to reflect the proportional representation of cell types in the subset as in the original dataset.
ncells: The target total number of cells for the subsetted object. Defaults to 1000. This parameter dictates the overall size of the subsetted SingleCellExperiment.
notEnough: Defines the behavior when there are not enough cells of a certain type to meet the subsetting criteria: "remove" to exclude that cell type entirely, or "asis" to include all available cells of that type despite falling short of the target number.
seed: An optional numeric value to set the random seed for reproducibility of the subsetting process. This ensures that the subset selection can be replicated in future analyses.

Value

Returns a subsetted SingleCellExperiment object. The resulting object reflects the targeted distribution and number of cells as specified by the input parameters. It enables focused analysis on a more manageable dataset while preserving the desired representation of cell types.