sample.genes.dep: Simulate Observations Given the Covariance Matrix for Normal...

Description Usage Arguments Details Value Author(s) Examples

Description

simulate a sample for genes given the covariance matrix for normal distribution

Usage

1
2
sample.genes.dep(n, delta, cov.ctrl, var.ratio)
sample.genes.dep.grp(n, delta, cov.ctrl, var.ratio)

Arguments

n

number of replicates for each observation(gene) for each of control/treatment group

delta

effect size for sample size calculation the minimum distance from zero for expression values of genes coming from the true alternative hypothesis group

cov.ctrl

covariance matrix for control group

var.ratio

a positive number for the ration of variance between treatment and control

Details

sample.genes.dep allows any size covariance matrix, while sample.genes.dep.grp assumes that the covariance matrix is 50x50.

The number of observation in the generated sample depends on the size of covariance matrix, the replicates for each observation for each control/ treatment group is specified by n

Value

returns a matrix with row number corresponding to the size of the covariance matrix and with 2*n columns: the first n columns being samples from control group and last n columns being from treatment group

Author(s)

Peng Liu peng_liu\@groton.pfizer.com

Examples

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
library(MASS)
# set.seed(100)
n <- 1000; delta <- 50; var.ratio <- 1;
cov.ctrl <- diag(rep(0.4,4))
cov.ctrl[1,1] <- 1
cov.ctrl[1,2] <- cov.ctrl[2,1] <- 0.15; # from test.sampling.cov

test.sample <- sample.genes.dep(n, delta, cov.ctrl, var.ratio)

## Not run: 

ndim <- dim(test.sample);
var.r <- rep(0, ndim[1])

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

generated.cov <- cov(t(test.sample[,1:n]), t(test.sample[,1:n]))


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

cat(" ------------------------------------------ \n",
    "number of genes (rows) :\t", length(cov.ctrl[1,]), "\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",
    "covariance matrix should be: \n")
print(cov.ctrl)

cat("the covariance matrix for the generated sample is: \n")
print(generated.cov)
cat("\n------------------------------------------ \n")

## End(Not run)	

warnes/exp.ssize documentation built on May 28, 2017, 12:58 a.m.