summarize_precision_recall: Compute annotation performance of cell types based on marker...

In gillislab/MetaMarkers: Extract robust markers from multiple datasets

Description

Compute annotation performance of cell types based on marker sets.

Usage

summarize_precision_recall(scores, true_labels, score_threshold)

Arguments

scores	A marker score matrix as returned by score_cells or compute_marker_enrichment.
true_labels	Ground truth labels for the target cells.
score_threshold	Numerical vector containing thresholds above which a cell should be annotated to a cell type.

Value

A data.frame where each row is a combination of a marker set, a ground truth cell type and a threshold. Marker scores are used as the predictor, the ground truth cell type defines a binary class (do cells belong to the given cell type?), and the threshold decides which cells are picked as positives or negatives. The performance is summarized with the following statistics: precision, recall and false positive rate.

gillislab/MetaMarkers documentation built on April 24, 2021, 9:25 p.m.