Spline-Based Drift Correction

step_measure_drift_spline() creates a specification of a recipe step that corrects for signal drift using smoothing splines fit to QC samples. This offers more flexibility than linear correction while being more stable than LOESS for sparse QC data.

Usage

step_measure_drift_spline(
  recipe,
  ...,
  run_order_col = "run_order",
  sample_type_col = "sample_type",
  qc_type = "qc",
  apply_to = c("all", "unknown"),
  df = NULL,
  spar = NULL,
  min_qc = 4,
  role = NA,
  trained = FALSE,
  skip = FALSE,
  id = recipes::rand_id("measure_drift_spline")
)

Arguments

recipe

A recipe object.

...

One or more selector functions to choose feature columns. For feature-level data, select the numeric response columns. For curve-level data with .measures, leave empty to apply to all locations.

run_order_col

Name of the column containing run order (injection sequence). Must be numeric/integer.

sample_type_col

Name of the column containing sample type.

qc_type

Value(s) in sample_type_col that identify QC samples to use for drift modeling. Default is "qc".

apply_to

Which samples to apply correction to:

• "all" (default): Correct all samples

• "unknown": Only correct unknown samples

df

Degrees of freedom for the smoothing spline. Default is NULL, which uses cross-validation to select optimal df. Lower values = smoother.

spar

Smoothing parameter (alternative to df). If NULL (default), cross-validation is used.

min_qc

Minimum number of QC samples required. Default is 5.

role

Not used by this step.

trained

Logical indicating if the step has been trained.

skip

Logical. Should the step be skipped when baking?

id

Unique step identifier.

Value

An updated recipe with the new step added.

Details

How It Works

Uses stats::smooth.spline() to fit a flexible curve through QC responses. The spline automatically adapts to the data complexity when df is not specified.

Comparison with Other Methods

Method	Best For	Min QC Samples
Linear	Simple linear drift	3
Spline	Moderate non-linearity	4+
LOESS	Complex patterns	5+

See also

step_measure_drift_linear() for linear correction, step_measure_drift_qc_loess() for LOESS-based correction.

Other drift-correction: step_measure_drift_linear(), step_measure_drift_qc_loess(), step_measure_qc_bracket()

Examples

library(recipes)

# Data with non-linear drift
set.seed(123)
data <- data.frame(
  sample_id = paste0("S", 1:30),
  sample_type = rep(c("qc", "unknown", "unknown", "unknown", "unknown", "qc"), 5),
  run_order = 1:30,
  feature1 = 100 + sin((1:30) / 5) * 10 + rnorm(30, sd = 2)
)

rec <- recipe(~ ., data = data) |>
  update_role(sample_id, new_role = "id") |>
  step_measure_drift_spline(feature1) |>
  prep()

corrected <- bake(rec, new_data = NULL)