Embedding-based palette with smooth gradients — roi_colors

Project ROI features to two dimensions (UMAP or PCA) and map polar angle to hue, yielding palettes that look globally structured.

Usage

roi_colors_embedding(
  rois,
  id_col = "roi",
  feature_cols = c("x", "y", "z"),
  hemi_col = NULL,
  method = c("umap", "pca"),
  C_range = c(40, 90),
  L = 65,
  seed = 1
)

Arguments

rois: Tibble with one ROI per row containing `roi`, `x`, `y`, and `z`.
id_col: Column containing ROI IDs.
feature_cols: Character vector of numeric or categorical columns that will be embedded (passed to `stats::model.matrix()`).
hemi_col: Optional hemisphere column used for the final light/dark adjustment.
method: Either `"umap"` (requires the `uwot` package) or `"pca"`.
C_range: Chroma range mapped from radial distance in the embedding.
L: Lightness value used for all ROIs before hemispheric adjustments.
seed: Random seed for reproducibility.

Value

Tibble with ROI IDs and colours.

Examples

# \donttest{
rois <- data.frame(
  roi = 1:10, network = rep(c("Vis", "DMN"), 5),
  x = runif(10), y = runif(10), z = runif(10)
)
pal <- roi_colors_embedding(rois, feature_cols = c("x","y","z"),
                             method = "pca")
# }