sample.hyp.ind: Function to Sample Observations(gene expressions) from...

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

Description

Function to Sample Observations(gene expressions) from Independent Normal Distribution with Simulated sd

Usage

sample.null.ind(ngenes.null.ind, n, sd.vector, var.ratio)
sample.alt.ind(ngenes.alt.ind, n, sd.vector, var.ratio, delta)

Arguments

ngenes.null.ind	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
sd.vector	vector of sd to sample from for controls
var.ratio	ratio of variance between treatment/control
ngenes.alt.ind	an integer number for the number of observations (gene expression values) to simulate for genes coming from true alternative hypothesis
delta	effect size, absolute difference of the means between genes from true alternative and true null nypothesis groups

Value

returns a matrix with ngenes.null.ind(or ngenes.alt.ind) 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)

Author(s)

Peng Liu peng\[e-mail]

Examples

n.null.ind <- 50; n <- 500; var.ratio <- 1;

test.null.ind <- sample.null.ind(n.null.ind , n, var.ratio)

## Not run: 

ndim <- dim(test.null.ind);
ssd.ctrl<-rep(0, ndim[1])
var.r <- rep(0, ndim[1])
k <- ceiling(n.null.ind * runif(1,0,1))

for ( i in c(1:ndim[1]))
  {ssd.ctrl[i]  <- sd(test.null.ind[i,1:n])
    var.r[i] <- var(test.null.ind[i,(n+1):(2*n)]) / var(test.null.ind[i,1:n])}

cat("comparison between the desired sample and the generated sample: \n")

cat(" ------------------------------------------ \n",
    "number of genes (rows) :\t", n.null.ind, "\t", ndim[1], "\n",
    "number of obs for each group: \t", n, "\t", ndim[2]/2, "\n",
    "variance ratio (trt/ctrl) is: \t", var.ratio, "\t", mean(var.r), "\n",
    "variances for ctrl are generated randomly from inverse gamma distrn. \n",
    "please check histogram of sd.\n",
    "------------------------------------------ \n")

### Histogram of the standard deviations generated

hist(ssd.ctrl,n=20, col="cyan", border="blue", main="",
     xlab="Standard Deviation (for simulated data)")
dens <- density(ssd.ctrl)
lines(dens$x, dens$y*par("usr")[4]/max(dens$y),col="red",lwd=2)

title("Histogram of Standard Deviations for generated sample")

## End(Not run) 

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