orQA (version 0.2.1)

e2test: Perform permutation Barlow's test on a dataset

Description

This function computes Barlow's test for each line of a given matrix. The global null distribution is computed using permutation. FWE control is provided by the maxT procedure.

Usage

e2test(data,g,B,rep=rep(1,length(g)))

Arguments

data

a numeric matrix, for the lines of which we want to test the null of no trend across groups

g

integer vector with group labels specified as ordered integers ranging from 1 to n where n is the number of ordered categories

B

number of permutations or a permutation matrix

rep

integer vector with group labels ranging from 1 to m where m is the number of independent samples (e.g. individuals). The ordering is not important.

Value

adj

vector with FWE adjusted p.values for each null hypothesis

raw

vector with raw p.values

dir

vector with the direction of each decision

Details

e2test takes a matrix for each line of which the null hypotheses of no trend across ordered groups is tested using a permutation test based on Barlow's E2 statistic. By permuting only samples within independent entities of the experimental design, random factors such as technical replicates from the same sample material can be incorporated. Multiple testing control is provided by the maxT procedure.

References

Klinglmueller, F., Tuechler, T., Posch, M. (2010) "Cross Platform Comparison Of Microarray Data Using Order Restricted Inference" Under Review Barlow, R. E., Bartholomew, D. J., Bremner, J. M., and Brunk, H. D. (1972) "Statistical inference under order restrictions"; Wiley, London.

Robertson, T., Wright,F. T. and Dykstra, R. L. (1988) "Order Restricted Statistical Inference"; Wiley, New York.

Examples

data <- matrix(rnorm(7200),nc=72)
groups <- rep(1:4,each=18)
ind <- rep(rep(1:3,each=6),4)
out <- e2test(data,groups,B=1000,rep=ind)
sum(out$adj<.05)
data2 <- data+matrix(rep(groups,nrow(data)),nr=nrow(data),byrow=TRUE)
out2 <- e2test(data2,groups,B=1000,rep=ind)
sum(out2$adj<.05)