In ayurovsky/rnaGinesis: Simulate gene expression data

knitr::opts_chunk$set(echo = TRUE)

R Markdown

Creates simulated data and saves expression matrix, trueH, trueW, and unmixed data to exp1.RData

Creates simulated data with 3 compartments, sample size = 50, noise_setting = TBD

init_info <- calculate_initial_mu_and_A()
mu <- init_info[[1]]
A <- init_info[[2]]

# no noise = 0; medium noise = 5; high_noise = 10;
Samplesize = 50
scaleFactor = rep(1000, 2)
noise_setting = 0
d.params = c("Tumor"   = .3,
             "Stromal" = .5,
             "Normal"  = .2)
sim.data <- Complete_simulation(A_tumor = A,
                               mu_tumor = mu, 
                               Samplesize = Samplesize,
                               scaleFactor = scaleFactor,
                               d.params = d.params,
                               noise_setting = noise_setting,
                               seed = 1234)

Obtains expression matrix with mixture data, trueW, trueH, signature matrices

ex <- sim.data[[1]]
meanW <- sim.data[[2]]
trueH <- sim.data[[3]]
trueW <- sim.data[[4]]

Dataset_Simulation <- list()
Dataset_Simulation$ex <- ex
Dataset_Simulation$trueW <- trueW
Dataset_Simulation$trueH <- trueH
Dataset_Simulation$unmixed <- meanW

# use the simulated H, but for W, make all samples identical
Dataset_Simulation <- list()
# make H
Dataset_Simulation$trueH <- trueH
# make unmixed
Dataset_Simulation$unmixed <- cbind(meanW[,1], sample(meanW[,1]),  sample(meanW[,1]))
colnames(Dataset_Simulation$unmixed) <- colnames(meanW)
rownames(Dataset_Simulation$unmixed) <- rownames(meanW)
# make trueW, same as unmixed
Dataset_Simulation$trueW <- lapply(1:50, function(X) Dataset_Simulation$unmixed)
# make expression matrix
Dataset_Simulation$ex <- matrix(, nrow=5000, ncol=0)
for (i in 1:50) {
  Dataset_Simulation$ex <- cbind(Dataset_Simulation$ex, as.vector(Dataset_Simulation$unmixed %*% Dataset_Simulation$trueH[,i]))
}
colnames(Dataset_Simulation$ex) <- colnames(ex)
rownames(Dataset_Simulation$ex) <- rownames(ex)

save(Dataset_Simulation, file = "~/rnaGinesis/data/Dataset_Simulation.RData")

ayurovsky/rnaGinesis documentation built on Dec. 19, 2021, 6:36 a.m.