Weighted Centered Log-Ratio Transformation — weighted_clr

Computes a weighted within-sample CLR transform. The weighted center uses down-weighted hemolysis-sensitive features, and the transformed coordinates can optionally be shrunk by the same weights before classification.

Usage

weighted_clr_transform(
  x,
  weights = NULL,
  sensitivity = .paper1_hemolysis_coefficients(),
  input_scale = c("log", "raw"),
  pseudocount = 1e-08,
  shrink_features = TRUE,
  feature_names = NULL
)

Arguments

x: Numeric vector or matrix with samples in rows and features in columns.
weights: Optional positive numeric vector of feature weights. If named, names are matched to `feature_names`.
sensitivity: Optional named positive numeric vector of hemolysis coefficients. Used only when `weights = NULL`.
input_scale: Either `"log"` (default) or `"raw"`.
pseudocount: Small positive constant added before logging raw inputs.
shrink_features: Logical; if `TRUE`, multiply the centered coordinates by their feature weights after centering.
feature_names: Optional feature names.

Value

Numeric vector or matrix of weighted CLR features.