[BioC] GO annotation inconsistency

[Daniel Gatti]

O/S: Windows XP
R: 2.3.1
Bioconductor: 1.8

I'm trying to get a list of all probes in a given GO category.  In the 
Bioconductor annotation libraries there are mapping from GO category to 
probe ID and from probe ID to GO category.  I'm finding that they do not 
match in terms of annotation.  Here's a sample script:

 library(hgu95av2)
 library(GO)
 
 # Get list of probe -> GO mappings.
 hgu95av2GO.list = as.list(hgu95av2GO)
 hgu95av2GO.list = lapply(hgu95av2GO.list, names)
 
 # Work with GO category 7031.
 GO.7031.probes = unique(get("GO:0007031", hgu95av2GO2ALLPROBES))
 length(GO.7031.probes)
[1] 16
 probe2GO.7031 = hgu95av2GO.list[match(GO.7031.probes, 
names(hgu95av2GO.list))]
 length(grep("GO:0007031", probe2GO.7031))
[1] 11

Note that the GO -> probe list gives me 16 probes in category 7031 while 
the probe -> GO list gives me 11 probes.  This happens for a lot of 
categories.  Am I missing some key concept or is there something else 
going on?

Thanks,
Dan Gatti
UNC-CH