compare_surrogate_estimates: Perform a comparison of the surrogate estimators demonstrated...

In elsayed-lab/hpgltools: A pile of (hopefully) useful R functions

Perform a comparison of the surrogate estimators demonstrated by Jeff Leek.

Description

This is entirely derivative, but seeks to provide similar estimates for one's own actual data and catch corner cases not taken into account in that document (for example if the estimators don't converge on a surrogate variable). This will attempt each of the surrogate estimators described by Leek: pca, sva supervised, sva unsupervised, ruv supervised, ruv residuals, ruv empirical. Upon completion it will perform the same limma expression analysis and plot the ranked t statistics as well as a correlation plot making use of the extracted estimators against condition/batch/whatever else. Finally, it does the same ranking plot against a linear fitting Leek performed and returns the whole pile of information as a list.

Usage

compare_surrogate_estimates(
  expt,
  extra_factors = NULL,
  filter_it = TRUE,
  filter_type = TRUE,
  do_catplots = FALSE,
  surrogates = "be",
  ...
)

Arguments

expt	Experiment containing a design and other information.
extra_factors	Character list of extra factors which may be included in the final plot of the data.
filter_it	Most of the time these surrogate methods get mad if there are 0s in the data. Filter it?
filter_type	Type of filter to use when filtering the input data.
do_catplots	Include the catplots? They don't make a lot of sense yet, so probably no.
surrogates	Use 'be' or 'leek' surrogate estimates, or choose a number.
...	Extra arguments when filtering.

Value

List of the results.

See Also

[normalize_expt()] [plot_pca()] [all_adjuster()] [corrplot] [ffpe]

elsayed-lab/hpgltools documentation built on May 9, 2024, 5:02 a.m.