Score cross-technology harmonized rCLR

Usage

score_cross_tech_harmonized_rclr(
  expr,
  meta,
  panel,
  technology_col = "technology",
  fit = NULL,
  train_expr = NULL,
  train_meta = NULL,
  required_technologies = NULL,
  min_detection_rate = 0.05,
  n_anchor = 20L,
  pseudocount = NULL
)

Arguments

expr: Numeric matrix, samples x features.
meta: Data frame aligned to `expr`.
panel: Character vector of panel features.
technology_col: Column identifying technology class.
fit: Optional fit from `fit_tech_stratified_rclr()`.
train_expr: Optional training matrix when `fit` is NULL.
train_meta: Optional training metadata when `fit` is NULL.
required_technologies: Technology classes required for selection. `NULL` (default) uses the known classes present in training metadata after alias mapping, excluding `unknown`. Fewer than two known classes raises `singlesample_input_error` with code `INSUFFICIENT_TECHNOLOGY_CLASSES`. An explicit vector retains the requested classes, including absent classes (which prevent anchor selection); an explicit single class remains allowed.
min_detection_rate: Minimum per-technology detection rate.
n_anchor: Maximum number of anchors to select.
pseudocount: Optional additive pseudocount before log transform.

Value

Numeric score vector.