[BioC] RP package

[fhong@salk.edu]

Hi Claire,

> Hi,
> I have been investigating using the RP package for data anaylsis (the
vignette was really helpful, thanks!) and I have a couple of points I
was unsure of. First of all I have 2 probesets (212280_x_at and
> 1557984_s_at) that appear once in the list of upregulated probesets and
once in the list of down-regulated probesets, why does this happen?

I need a little more information here. Did you mean that you run Rank
Product using different random number generator, these two genes appear to
be upregulated once and down-regulated once?
Since rank product searches for up-regulated genes and down-regulated
genes  separately, this would be possible when the behavior of this gene
is really inconsistent, meaning having really large fold change from some
replicates and really small food-change under some other replicates.

I would suggest that you check the average fold-change output from
topGene, and the Orirank output from RP ( which is accessed by
rp.out$Orirank[[1]]: fro the up-regulation search, rp.out$Orirank[[2]] for
the down-regulation search) to double check the identified genes.



> Also
> for 1 probeset (1557984_s_at), fold changes are greater than 1 in the
downregulated list but less than 1 in the upregulated list.
This is strange since the fold changes are computed once, it should be the
same value no matter which list this probeset appears?


> Any ideas why this is happening? I have included the relevant sections
from $Table1 and $Table2 in the attached file
If you can send me the data set, rma.eset at exprs, I might be able to check
it .

Thanks.
Fangxin




> library(RankProd)
> data.classes <- c(0,0,1,0,0,1,0,0,1,0,0,1,0,0,1,0,0,1)
> rp.out <-
> RP(rma.eset at exprs,data.classes,gene.names=rownames(rma.eset at exprs),rand123)
output <-
> rbind(topGene(rp.out,cutoff=0.05)$Table1,topGene(rp.out,cutoff=0.05)$Tab
le2)
> annot <-
> read.csv("/users/cwilson/ying/annotAug05.txt",header=T,sep="\t") tmp <-
annot[,2:3]
> rownames(tmp) <- annot[,1]
> annot <- tmp
> output <- cbind(annot[output[,1],],output)
> write.table(x=output,file="rp_out.xls",quote=F,sep="\t")
> --
> Claire Wilson
> Bioinformatics Group
> Paterson Institute for Cancer Research
> Christie Hospital NHS Trust
> Wilmslow Road
> Manchester
> M20 4BX
> tel: +44 (0)161 446 8218
> url: http://bioinf.picr.man.ac.uk/bioinf/cwilson.jsp
> <http://bioinf.picr.man.ac.uk/bioinf/cwilson.jsp>
> --------------------------------------------------------
> This email is confidential and intended solely for the use...{{dropped}}