callHarman: Wrapper function to call the shared C/C++ library code

In Harman: The removal of batch effects from datasets using a PCA and constrained optimisation based technique

Description

This wrapper should probably not be addressed directly except for debugging. Instead use harman. Input of PCA scores and the experiment structure (treatments and batches) and returns a batch corrected version of the PCA scores matrix

Usage

.callHarman(pc_data_scores, group, limit, numrepeats, randseed, forceRand,
  printInfo)

Arguments

pc_data_scores	2D NumericMatrix of PCA scores data (from the prcomp$x slot), rows = samples, cols = PC scores
group	The structure of the experiment, consisting of batch numbers and treatment numbers forming 2 rows or columns (HarmanMain works out which). Each entry for a sample describes what batch it came from and what treatment it was given. Has to be integer formated data.
limit	A double precsion value indicating the limit of confidence in which to stop removing a batch effect
numrepeats	The number of repeats in which to run the simulated batch mean distribution estimator. Probably should be greater than 100,000.
randseed	Random seed to pass to the random number generator (0 for use default from system time)
forceRand	Force algorithm
printInfo	Print update information to screen

Value

SEXP R list: scores.corrected = harman_res_list["corrected_scores"] correction = harman_res_list["correction"] confidence = harman_res_list["confidence"]

Note

A data matrix with samples in columns must be transposed before PCA analysis and these scores in turn are tweaked a little before handing over to .callHarman. See the example below.

Harman documentation built on Nov. 8, 2020, 7:50 p.m.