generateMixedData: Generate mixed data with or without noise

In ctlab/LinSeed: Linear Subpsace identification to solve complete gene expression deconvolution problem

Description

Generates mixed data with or without noise under assumption of linear model

Usage

generateMixedData(samples = 40, genes = 12000, cellTypes = 3,
  bias = NULL, spearmanThreshold = 0.5, pureGenes = 0,
  noiseDeviation = 0, sampleLogMean = 4, sampleLogSd = 3,
  removeAngles = F, cutCoef = 0.85, removeBorders = F,
  borderShift = 0.25)

Arguments

samples	number of samples
genes	number of genes
cellTypes	number of cell types
bias	if not null bias should be a number in between 0 and 1, one of cell types presented in mix will be more abundunt and one cell type will be less abundunt
spearmanThreshold	Spearman treshold that has to be between generated samples
pureGenes	number of genes which are not noisy
noiseDeviation	standart deviation of normally distributed noise value
sampleLogMean	average log expression of pure samples
sampleLogSd	standard deviation of log expression of pure samples
removeAngles	if TRUE there will be no signature genes in the mix (simplex corners will be removed)
cutCoef	if removeAngles == T how much to cut angles
removeBorders	if TRUE where will be no genes close to simplex border
borderShift	number in between 0 and 1, every value in basis is guaranteed to be at least borderShift

ctlab/LinSeed documentation built on Aug. 9, 2019, 4:33 p.m.