simu.sample: Simulate a sample with normal distribution

Description Usage Arguments Value Author(s) Examples

Description

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

Usage

1
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\[email protected]

Examples

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
16
17
18
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 28, 2017, 12:58 a.m.