Apply frozen hemolysis-RR correction to new samples — apply_hemolysis

Subtracts the predicted hemolysis (and optional platelet) nuisance contribution from one or more test samples using the frozen per-feature regression coefficients from fit_hemolysis_rr.

Corrections are clipped to \(\pm\)clip_max log-units to prevent over-correction. If report_oob = TRUE (default), an oob attribute flags samples whose hemolysis / platelet scores fall outside the training range.

Usage

apply_hemolysis_rr(
  x,
  fit,
  hemo_score,
  platelet_score = NULL,
  report_oob = TRUE
)

Arguments

x: Numeric vector (single sample) or matrix (samples \(\times\) features).
fit: Object of class hemolysis_rr_fit from fit_hemolysis_rr.
hemo_score: Numeric vector (length = number of samples).
platelet_score: Optional numeric vector (same length as hemo_score).
report_oob: Logical. If TRUE, attaches an oob attribute flagging out-of-training-range samples. Default TRUE.

Value

Same shape as x. Hemolysis-corrected log-expression values, with attributes oob (if report_oob) and fit_summary.