Apply frozen RUV correction to new samples

Removes the unwanted-variation component estimated by fit_frozen_ruv from one or more new samples. The latent factors are approximated from the frozen pseudo-inverse learned during training.

Usage

apply_frozen_ruv(x, fit)

Arguments

x

Numeric vector (single sample) or matrix (samples \(\times\) features).

fit

A frozen_ruv_fit object from fit_frozen_ruv.

Value

Same shape as x, with the unwanted-variation component removed and training feature means re-added. The attribute fit_summary summarises the fit parameters.