Score with the mixed-effects scorer — score_mixed_effects

If `fit` is supplied, this function predicts from the existing fit and does not read disease labels from `meta`. If `fit` is NULL, `train_expr`, `train_meta`, and `outcome_col` are required and the model is fit on that training pool before prediction.

Usage

score_mixed_effects_scorer(
  expr,
  meta,
  panel,
  cohort_col = "cohort",
  block_col = "provenance_block",
  fit = NULL,
  train_expr = NULL,
  train_meta = NULL,
  outcome_col = "disease",
  pseudocount = NULL
)

Arguments

expr: Numeric matrix, test samples x features.
meta: Data frame aligned to `expr`.
panel: Character vector of panel features.
cohort_col: Cohort column.
block_col: Provenance-block column.
fit: Optional fit from `fit_mixed_effects_scorer()`.
train_expr: Optional training expression matrix when `fit` is NULL.
train_meta: Optional training metadata when `fit` is NULL.
outcome_col: Binary disease label column in `train_meta`.
pseudocount: Optional additive pseudocount before log transform.

Value

Numeric score vector.