Score with provenance-block fixed-effect rCLR centering

Usage

score_block_fe_rclr(
  expr,
  meta,
  panel,
  cohort_col = "cohort",
  block_col = "provenance_block",
  fit = NULL,
  pseudocount = NULL
)

Arguments

expr: Numeric matrix, samples x features.
meta: Data frame aligned to `expr`.
panel: Character vector of panel features.
cohort_col: Column in `meta` identifying cohorts; accepted for a common scorer signature but not used by this method.
block_col: Column in `meta` identifying provenance blocks.
fit: Optional object from `fit_block_fe_rclr()`.
pseudocount: Optional additive pseudocount before log transform.

Value

Numeric score vector, one value per sample. Attribute `fallback_samples` counts samples assigned to the training global center because their block was absent from the fit.