Skip to contents

Apply a pipeline independently to each group of a SingleCells object

apply_pipeline_per_group.Rd

Splits object by group_col, applies pipeline to each subset, and returns a named list of processed SingleCellsSubsets. Useful for per-sample / per-cell-type re-analysis where the same chain (HVG, PCA, neighbours, clusters, ...) is run on each group, e.g. sample-pure metacell generation followed by an external merge.

Usage

apply_pipeline_per_group(
  pipeline,
  object,
  group_col,
  groups = NULL,
  progress = FALSE
)

Arguments

pipeline

ScPipeline.

object

SingleCells.

group_col

String. Column in obs used to split.

groups

Optional character vector. Restrict to these group values; if NULL, all unique values of group_col are used.

progress

Boolean. Shall big progress messages be printed to the console. Defaults to FALSE.

Value

Named list of processed objects, names being the group values. Usually SingleCellsSubset, or MetaCells if the pipeline ends on step_metacells_sc(), in which case merge_meta_cells() puts them back together.

Examples

# the same chain re-run inside each cell type
sc <- demo_single_cells(prepped = FALSE)
p <- sc_pipeline() %>>% step_hvg_sc(hvg_no = 20L, .verbose = FALSE)
res <- apply_pipeline_per_group(p, sc, group_col = "cell_grp")
names(res)
#> [1] "cell_type_1" "cell_type_2" "cell_type_3"

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