sample.hyp.dep: Generate observations from each hypothesis (null/alternative)...

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

Description

Generate observations from each hypothesis (null/alternative) group based on dependent normal distribution

Usage

sample.null.dep(ngenes.null.dep, n, var.ratio, cov.matrix,
                ngenes.matrix, distrn = "normal")

sample.alt.dep(ngenes.alt.dep, n, var.ratio, cov.matrix, ngenes.matrix,
               delta, distrn = "normal")

Arguments

ngenes.null.dep	an integer number for the number of observations (gene expression values) to simulate for genes coming from true null hypothesis
n	number of replicates for each group of control/treatment
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
distrn	always being normal
ngenes.alt.dep	an integer number for the number of observations (gene expression values) to simulate for genes coming from true alternative hypothesis
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.***.dep 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

Author(s)

Peng Liu peng\[e-mail]

Examples

library(MASS)
ngenes.null.dep <- ngenes.alt.dep <- 10
n <- 1000; delta <- 50; var.ratio <- 1;

ngenes.matrix <- 7; cov.matrix <- list(ngenes.matrix);

for ( i in c(1:ngenes.matrix)) {
    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.null.dep <- sample.null.dep(ngenes.null.dep , n, var.ratio,
      cov.matrix, ngenes.matrix)
 test.sample.alt.dep <- sample.alt.dep(ngenes.alt.dep , n, var.ratio,
      cov.matrix, ngenes.matrix, delta)

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