[R] FDR for hypergeometric tests

darteta001 at ikasle.ehu.es darteta001 at ikasle.ehu.es
Tue Jan 15 16:32:30 CET 2008


Dear list,

I have performed several tests for the hypergeometric distribution 
using phyper() for some gene annotation categories as follows

>phyper(26,830,31042,337, lower.tail=F)
>phyper(16,387,31042,337, lower.tail=F)
 .
 .
 .

I am only running some selected categories but I would like to correct 
this value for multiple testing since I 
have 3121 possible tests according to 3121 annotation categories. I 
think that for 
the case of Bonferroni corrections, I should correct this output by 
the total number of possible tests conducted (3121):

>phyper(26,830,31042,337, lower.tail=F)*3121
>phyper(16,387,31042,337, lower.tail=F)*3121

But I would like to perform FDR, i.e. Benjamini correction. I know 
that multtest package has the function mt.rawp2adjp() but I would need 
to save all the p.values for the 3121 in a vector and use this as a 
parameter of 
the function. Is there another way of computing this adjusted p.value 
for only some of the categories?

Thanks for your help,

David

(R version 2.6.0)
(Windows OS)




More information about the R-help mailing list