Agreement rates between two label columns

Computes per-label agreement between two label columns in `colData(spe)`. For each label (union of both columns), agreement is defined as the number of cells where both labels match that value divided by the number of cells where either column uses that value (Jaccard-style overlap).

Usage

label_agreement_rates(spe, label_col1, label_col2, drop_na = TRUE)

Arguments

spe

A `SpatialExperiment`/`SingleCellExperiment`.

label_col1

First label column name.

label_col2

Second label column name.

drop_na

Logical; if TRUE, rows with NA in either label are dropped. Default TRUE.

Value

A data frame with columns `label`, `match_n`, `union_n`, and `agreement`.

Examples

data("cytoGateR_example", package = "cytoGateR")
label_agreement_rates(
  cytoGateR_example,
  "cell_type_hard",
  "knn_label"
)
#>          label match_n union_n   agreement
#> 1        Bcell     280     647 0.432766615
#> 2      BnTcell    1279    2602 0.491544965
#> 3          CD4       5     891 0.005611672
#> 4          CD8     138     660 0.209090909
#> 5      Myeloid     450    1132 0.397526502
#> 6   Neutrophil     314    1036 0.303088803
#> 7  Plasma_cell      26     157 0.165605096
#> 8       Stroma      80    1051 0.076117983
#> 9         Treg     190     731 0.259917921
#> 10       Tumor      10    1242 0.008051530
#> 11  Unassigned       0     145 0.000000000
#> 12   Uncertain       0    1342 0.000000000