Assign cell labels using broad-marker enrichment

Assigns broad labels to individual cells using aggregated log-normalised expression across small curated marker sets for each broad category.

Usage

annotate_broad_cells(
  sce,
  marker_config_key = "marker_config",
  label_col = "broad_enrichment",
  assay_name = "logcounts",
  aggregate_fun = c("sum", "mean", "median"),
  other_label = "other"
)

Arguments

sce

A SingleCellExperiment containing a logcounts assay.

marker_config_key

Character. Metadata entry containing a named list of validated marker definitions. Defaults to "marker_config".

label_col

Character scalar. Name of the output column to create in colData(sce). Defaults to "broad_enrichment".

assay_name

Character scalar. Assay to use for expression values. Defaults to "logcounts".

aggregate_fun

Character scalar specifying how marker expression should be aggregated within each category. One of "sum", "mean", or "median". Defaults to "sum".

other_label

Character scalar. Label assigned when no category passes. Defaults to "other".

Value

The input SingleCellExperiment with:

colData(sce)[[label_col]]

A factor of per-cell broad labels, with other_label as the last level.

metadata(sce)$broad_cell_enrichment

A list containing the per-cell category score matrix, logical pass matrix, and assignment parameters.

Details

For each category, expression is aggregated across that category's marker genes using one of "sum", "mean", or "median". A category is considered to pass for a cell if the aggregated expression exceeds the category-specific expr_threshold. If no categories pass, the cell is labelled as other_label. If more than one category passes, the tie is resolved using the user-supplied priority, with lower numeric values treated as higher priority.

This function assumes that broad_config has already been validated in upstream preprocessing, including checks that all broad markers are present, marker sets do not overlap, and priorities are unique.