[BioC] affycoretools vennSelect problem

[Peadar Ó Gaora]

Hi Jim,

Thanks for getting back to me.

On Mon, 2008-06-23 at 09:18 -0400, James W. MacDonald wrote:
> Hi Peadar,
> 
> Peadar Ó Gaora wrote:
> > Hello BioCer's,
> > 
> > I've been using affycoretools, particularly limma2annaffy,for a while
> > now and love it.  These wrappers really make life simpler.
> > 
> > However (there was bound to be a however), in a recent experiment I've
> > had some problems with vennSelect.  The experiment has 4 groups and 4
> > time points, 5 replicates each.  There are then, a plethora of possible
> > contrasts of interest.  I've run one particular analysis with 34
> > contrasts and wanted to look at the unique (differentially expressed,
> > p<0.05) genes and intersections between some of them.  When I use
> > vennSelect to do this, it returns probesets with P values well above
> > 0.05.
> 
> If you have more than three contrasts, vennSelect() is supposed to error 
> out with a message - is that not happening?

vennSelect behaves exactly that way if I feed it more than 3 contrasts.
What I've been doing is subsetting the TestResults and contrast matrix
to the two contrasts I want to look at eg:

vennSelect(gcrma_Data, design, interday_res[,c("pregHi_d5vd7",
"pregLo_d5vd7")], interday.contrasts[,c("pregHi_d5vd7",
"pregLo_d5vd7")], interday_fit2, stat="tstat")

where pregHi_d5vd7 and pregLo_d5vd7 are contrasts of interest.

Having written this, I've just found that the problem was indeed passing
in the entire fit object.  Subsetting this as well corrects the issue.
I had tried and failed to do that before but I guess I had a typo in the
command.  Anyway, I just thought I should finish this out in case
anybody checking the archives in the future has a similar question.

so the correct call should have been: 

vennSelect(gcrma_Data, design, interday_res[,c("pregHi_d5vd7",
"pregLo_d5vd7")], interday.contrasts[,c("pregHi_d5vd7",
"pregLo_d5vd7")],interday_fit2[,c("pregHi_d5vd7", "pregLo_d5vd7")],
stat="tstat")


Go raibh míle maith agat Jim,

Peadar


> 
> > 
> > I can't see where it is picking these from as the lists from the
> > individual contrasts all look perfectly fine (P-value for all selected
> > genes < 0.05).  I have tried subsetting the TestResults and contrasts
> > matrices in the call to vennSelect.  I've also tried generating new ones
> > TestResults objext and contrast matrix which include only the
> > comparisons being analysed.  The fit object however contains the entire
> > dataset.  Could this be where things are going wrong?
> 
> I haven't made many changes to affycoretools for a while, but that 
> version of R/BioC is seriously ancient. The first step should be to 
> upgrade to R-2.7.1 and BioC 2.2.
> Then you should show the code you are using (hopefully simplified if 
> possible).
> 
> Best,
> 
> Jim
> 
> 
> 
> > 



> > 
> > Any help much appreciated.
> > 
> > Peadar
> > 
> > 
> >> sessionInfo()
> > R version 2.5.1 (2007-06-27)
> > ia64-unknown-linux-gnu
> > 
> > locale:
> > LC_CTYPE=en_IE.UTF-8;LC_NUMERIC=C;LC_TIME=en_IE.UTF-8;LC_COLLATE=en_IE.UTF-8;LC_MONETARY=en_IE.UTF-8;LC_MESSAGES=en_IE.UTF-8;LC_PAPER=en_IE.UTF-8;LC_NAME=C;LC_ADDRESS=C;LC_TELEPHONE=C;LC_MEASUREMENT=en_IE.UTF-8;LC_IDENTIFICATION=C
> > 
> > attached base packages:
> > [1] "splines"   "tools"     "stats"     "graphics"  "grDevices"
> > "datasets"
> > [7] "utils"     "methods"   "base"
> > 
> > other attached packages:
> > affycoretools       annaffy        xtable         gcrma   matchprobes
> >       "1.8.1"       "1.8.1"       "1.5-1"       "2.8.1"       "1.8.1"
> >       biomaRt         RCurl           XML       GOstats      Category
> >      "1.10.1"       "0.8-0"      "1.92-1"       "2.2.6"       "2.2.3"
> >        Matrix       lattice    genefilter      survival          KEGG
> >  "0.999375-2"     "0.15-11"      "1.14.1"        "2.32"      "1.16.1"
> >          RBGL      annotate            GO         graph         limma
> >      "1.12.0"      "1.14.1"      "1.16.0"      "1.14.2"      "2.10.5"
> >          affy        affyio       Biobase
> >      "1.14.2"       "1.4.1"      "1.14.1"
> > 
>