[BioC] FRD : package fdrame vs multtest
Nicolas Servant
Nicolas.Servant at curie.fr
Wed Nov 9 17:58:02 CET 2005
Hi all,
I would like to compare the two packages "multtest" and "frame" to
compare the two FDR function.
I test them on the Golub data, and perform the simple LSU (Linear Step
Procedure) procedure.
The both results are not exactly the same...
##multtest package
test<-mt.teststat(golub,golub.cl, test="t.equalvar")
pval.test <- 2*(pt(-abs(test),36))
resFDR<-mt.rawp2adjp(pval.test, proc=c("BH"))
adjp <- resFDR$adjp[order(resFDR$index),]
##fdrame
fdrame<-fdr.ma(golub,golub.cl,p.method="theoretic",fdr.adj="BH-LSU")
The fdrame package use the Student test, and i verified that the
fdram$dif data were the same that the mt.teststat$teststat.num data.
Therefore i suppose that the no adjusted pvalues are equally the same.
The first 10 results (pval fdrame adjp-multtest):
[1] "0.0170276669459720 0.201688145549621 0.068292673088116"
[2] "0.255228106990335 0.544842407507294 0.436736373767533"
[3] "0.913030948999123 0.959693978582492 0.956613126853133"
[4] "0.786674366368 0.841218391749495 0.879495599776024"
[5] "0.243146660088367 0.334140455294467 0.425855602715044"
[6] "0.281983317344268 0.376018736206403 0.466714155158842"
[7] "0.62418676693191 0.645068479616667 0.769314565346159"
[8] "0.716299487029608 0.74157426271768 0.83001509112318"
[9] "0.991113269533743 0.992994429086477 0.995026846116305"
[10] "0.928212225072347 0.943350294841212 0.96310489568757"
If anybody have an idea to explain how can i observe such a difference ...
Is it exactly the same procedures ?
Best regards,
Nicolas Servant
--
Nicolas Servant
Equipe Bioinformatique
Institut Curie
26, rue d'Ulm - 75248 Paris Cedex 05 - FRANCE
Email: Nicolas.Servant at curie.fr
Tel: 01 44 32 42 75
More information about the Bioconductor
mailing list