Apply Protected Spatial Priors to kNN Probabilities

Usage

calculate_spatial_prior_labels(
  spe,
  prob_mat,
  k_spatial = 50,
  lambda = 0.2,
  protect_threshold = 0.85,
  out_col = "knn_spatial_label"
)

Arguments

spe: SpatialExperiment object.
prob_mat: The full probability matrix [Cells x Types] from predict_unknown_with_knn.
k_spatial: Number of physical neighbors for the "Prior" (default 50).
lambda: The "Spatial Weight" (default 0.2).
protect_threshold: Confidence level above which spatial priors are ignored (default 0.85).
out_col: Column name for the new labels.

Value

The input `SpatialExperiment` with the spatially adjusted labels stored in `colData()` under `out_col`.

Examples

data("cytoGateR_example", package = "cytoGateR")
image_ids <- SummarizedExperiment::colData(cytoGateR_example)$image_name
spe <- cytoGateR_example[, image_ids == image_ids[1L]]
prob_cols <- grep(
  "^KNN_P_",
  names(SummarizedExperiment::colData(spe)),
  value = TRUE
)
prob_mat <- as.matrix(SummarizedExperiment::colData(spe)[, prob_cols])
rownames(prob_mat) <- colnames(spe)
colnames(prob_mat) <- sub("^KNN_P_", "", colnames(prob_mat))
spe <- calculate_spatial_prior_labels(
  spe,
  prob_mat,
  k_spatial = 3,
  out_col = "spatial_label"
)
#> Integrating spatial context (Protecting cells > 0.85 confidence)...
table(SummarizedExperiment::colData(spe)$spatial_label)
#> 
#>       Bcell     BnTcell         CD4         CD8     Myeloid  Neutrophil 
#>         714        1392         777         514         274          36 
#> Plasma_cell      Stroma        Treg       Tumor 
#>          40         398         135          83