[BioC] Questions about clusterization from the newbie

[James W. MacDonald]

tiv_ wrote:

> # strange, but mt.teststat(data_rma,data.c) its not working
> # without  writing and reading back, is it ok?

mt.teststat is expecting your data to be in a matrix or a data.frame, so 
you have to first extract using the exprs() function, e.g., 
mt.teststat(exprs(data_rma), data.c)

> 
> data.c<-c(0,0,1,1)
> teststat<-mt.teststat(x,data.c)
> rawp0<-2*(1-pnorm(abs(teststat)))

This is not correct. The p-values for a t-statistic are calculated based 
on the t-distribution rather than the normal distribution (which will 
make a big difference with only two degrees of freedom).

rawp0 <- 2*(1-pt(abs(teststat), df=2)


> procs<-c("Bonferroni")
> res<-mt.rawp2adjp(rawp0,procs)
> adjp<-res$adjp[order(res$index),]
> which<-mt.reject(adjp,0.01)$which[,2]
> results<-table2[which,2] #table2 is the annotation table
> saveText(results,"sorted_bonf.txt")
> 
> As a result I have a list of genes that changed their expression. Can
> you give me a hint how can I separate "up regulated" from "down
> regulated", must be smiting with t-statistics?

Maybe something like this:

sig.genes <- teststat[which]
ups <- sig.genes > 0
downs <- sig.genes < 0
up.genes <- results[ups]
down.genes <- results[downs]

-or-

results <- cbind(results, sig.genes)
results <- results[order(results[,2]),]

HTH,

Jim


-- 
James W. MacDonald
Affymetrix and cDNA Microarray Core
University of Michigan Cancer Center
1500 E. Medical Center Drive
7410 CCGC
Ann Arbor MI 48109