Score cluster marker genes against fine cell type marker panels — score_markers_against

For each cluster, identifies the relevant subset of the fine cell type marker panel based on the broad label encoded in the cluster name, then scores overlap using Fisher's exact test and overlap similarity.

Usage

score_markers_against_panel(top_markers, marker_panel, background_genes)

Arguments

top_markers: Named list. Output of extract_top_markers; each element must contain a top_n character vector.
marker_panel: Named list of named lists. Top-level names are broad cell-type categories (e.g. "immune", "vasculature", "other"); inner names are fine cell types; values are character vectors of marker genes.
background_genes: Character vector. Full set of genes in the dataset, used as the background population for Fisher's exact test.

Value

A data.frame with columns:

cluster_name: Cluster identifier from top_markers.
marker_set_name: Fine cell type name from the panel.
fisher_p: P-value from Fisher's exact test.
similarity: Overlap similarity score.

Rows are ordered by cluster_name then ascending fisher_p.

Details

Cluster names are expected to follow the convention "<broad>_sc<N>", e.g. "immune_sc1", "vasculature_sc2". The broad prefix is matched against the top-level names of marker_panel.

Edge case — collapsed broad labels: When all clusters share the same broad label (e.g. all assigned "other"), downstream subclustering produces names like "1_sc1", "2_sc2" whose prefix is numeric and does not resolve to a panel entry. In this situation the function tests against all marker sets in the panel.