[BioC] match probes: probesets with too few matches

[Margaret Gardiner-Garden]

Hi, I have been using the function combineAffyBatch (match probes library)
to combine the data for two mouse chips MOE430v2 (45101 probesets) and its
subset chip MOE430Av2 (22690 probesets).
According to Affymetrix the 22690 probesets in MOE430Av2 are identical in
MOE430v2.

I ran the following:
res <- combineAffyBatch(list(Data.430v2, Data.430Av2),c("mouse4302probe",
"mouse430a2probe"), newcdf="comb")
comb <- res$cdf


This gave a combination expression set with 22702 probesets (ie 12 more
probeset names than the subset chip MOE430Av2).

I looked to see the number of oligos matching in the different probesets
prs <- mget(ls(comb), comb,  ifnotfound=NA)
nrprobes <- sapply(prs, function(x) nrow(x))
table(nrprobes)


> table(nrprobes)
nrprobes
    1     2     3     4     5     8     9    10    11    20    21
    7     2     1     1     1     1     4    11 22631    40     3



The probeset names that were not present on MOE430v2 all had less than 8
oligos matching between the chip types.

I was wondering if anyone knew the best way to exclude such probesets in the
expression set, or how to avoid this problem in the first place.


Thanks and Regards
Dr Margaret Gardiner-Garden