simu.sample: Simulate a sample with normal distribution

In warnes/exp.ssize: Estimate Effectivenes of Microarry Sample Size Estimation via Simulation

Description

Simulate samples from normal distributions with specified proportion of observations from true alternative hypothesis, dependent groups, variance ratio for treatment/control group

Usage

simu.sample(ngenes, n, fractn.alt, fractn.dep, var.ratio, cov.matrix, ngenes.matrix, delta)

Arguments

ngenes	number of genes to simulate
n	number of replicates for each group of control/treatment
fractn.alt	the fraction of genes coming from true alternative hypothesis
fractn.dep	the fraction of genes coming from dependent group
var.ratio	ratio of variance between treatment/control
cov.matrix	a list of covariance matrices to sample the covariance from
ngenes.matrix	the number of covariance matrices in the list of cov.matrix
delta	effect size for sample size calculation the absolute distance from zero for expression values of genes coming from the true alternative hypothesis group

Value

returns a matrix with ngenes rows and 2*n columns with the first n columns being samples from control group and last n columns being from treatment group for each gene (row). each group of genes consists of subgroups of genes that are dependent and/or coming from true alternative hypothesis. the row order is: null.ind -> null.dep -> alt.ind -> alt.dep

Author(s)

Peng Liu peng\[e-mail]

Examples

library(MASS)
cov.matrix <- list(7);

for ( i in c(1:7)) {
    cov.matrix[[i]] <- diag(rep(i*0.1, i+1))
    # cov.matrix[[i]] <- diag(rep(i*0.1,4))
}

cov.matrix[[4]][1,1] <- cov.matrix[[5]][1,1] <- cov.matrix[[6]][1,1] <- 1
cov.matrix[[4]][2,1] <- cov.matrix[[4]][1,2] <- 0.15 # corr = 0.15 / (1*0.2) = 0.75
cov.matrix[[5]][2,1] <- cov.matrix[[5]][1,2] <- 0.5 # corr = 0.15 /(1*sqrt(0.5)) = 0.59
cov.matrix[[7]][1,1] <- 10
cov.matrix[[7]][2,1] <- cov.matrix[[7]][1,2] <- 0.6 # corr = 0.67
cov.matrix[[7]][3,1] <- cov.matrix[[7]][1,3] <- 0.5
cov.matrix[[7]][3,2] <- cov.matrix[[7]][2,3] <- 0.4

test.sample <- simu.sample(ngenes = 100, n = 6, fractn.alt = 0.8, fractn.dep = 0.2,
   var.ratio=1, cov.matrix, ngenes.matrix = 7, delta = 10)

warnes/exp.ssize documentation built on May 4, 2019, 12:59 a.m.