[BioC] Bug in probeNames()

[lgautier at altern.org]

> Hello list,
>
> I ran into the following particular problem using the Custom CDF-files
from Bioconductor 2.0. Please consider the following code:
>
> library(simpleaffy)
> x <- ReadAffy(cdfname="mm430mmentrezg")
> require(sprintf("%sprobe", cleancdfname(cdfName(x), addcdf=F)),
> character.only=T)
> ProbeNames_CustomCDF <-
> as.character(as.data.frame(mm430mmentrezgprobe)[,4])
>
> grep("66736_at", ProbeNames_CustomCDF, fixed=T)
> 173151 173152 173153 173154 173155 173156 173157
>
> However, when I do it the following way different indices are returned:
>
> grep("66736_at", probeNames(x), fixed=T)
> 172057 172058 172059 172060 172061 172062 172063
>
> I do not understand the reason for this! To get the order of the
basepairs
> (for this particular probe) I need to use the upper grep-command (I
extract the basepairs from the probes library), but to get the CORRECT
pm() and mm() intensities I need to use the lower grep-command. Why is
this? For the Affymetrix annotation libraries everything works fine (as I
> would expect).

It might just be working on Affymetrix libraries by chance.

To my knowledge corresponding indexes is not enforced in
the code: "probeNames" (that could have been rather named
"probesetNames") returns an ordered list of the keys in
the corresponding CDF environment (and freedom is given to
redefine CDF environments). If I remember right the data.frame
in the default probe package is read from a file provided by Affymetrix
(and the order of the probesets is the one found in the file).

The probe packages contain a data.frame and, although the function
getProbeDataAffy in the package matchprobes makes some checks by default
checks against the matching CDF, there are no explicit checks for matching
indexes at build-time, neither there are run-time consistency checks.

The problem you encounter can also be seen as a problem with the pairs
environment/probe packages you are using.

Hoping this helps,


Laurent



> Regards,
>
> Dr Philip de Groot
> Wageningen University
>
> _______________________________________________
> Bioconductor mailing list
> Bioconductor at stat.math.ethz.ch
> https://stat.ethz.ch/mailman/listinfo/bioconductor
> Search the archives:
> http://news.gmane.org/gmane.science.biology.informatics.conductor
>