[BioC] Multtest question...

[Joshua Betcher]

Here is a very contrived example:

x.temp <- rbind(c(4,4,4,4),
                        c(1,1,4,4),
                        c(1,0,3,4),
                        c(1,.99999,.99998,.99997),
                        c(1,.99999,.99998,.99998))
classlabels <- c(0,0,1,1)

t.test <- mt.teststat(x.temp, classlabel = classlabels, test= "t")
rawp0 <- 2 * (1 - pnorm(abs(t.test)))

mt.rawp2adjp(rawp0, "Bonderroni")


Output:

$adjp
             rawp Bonderroni
[1,] 0.000000e+00          0
[2,] 0.000000e+00          0
[3,] 2.209050e-05          0
[4,] 2.699796e-03          0
[5,] 4.677735e-03          0

$index
[1] 1 2 3 5 4


I am not too troubled by the results of the last two rows but the first two
seem to be concerning, the method does not seem to take into account zero
differences.  Am I correct in this account?  If so, how would one fix this.

Thanks
Josh

Software: WinXP R 1.5.0, Latest version of Multtest