[BioC] help with ath1121501

[Seth Falcon]

Hi Martin,

Martin Olivier <martinol at ensam.inra.fr> writes:
> I have (again) some problems to use the package ath1121501.
> I don't understand the difference between ath1121501G02PROBE and
> ath1121501GO2ALLPROBES...

Here's the description from the help pages of those two environments:

     ath1121501GO2PROBE is an R environment that provides mappings
     between GO identifiers and manufacturer identifiers.

     ath1121501GO2ALLPROBES is an R environment that provides mappings
     between a given GO identifier and all manufactuerer identifiers
     annotated at that GO term or one of its children in the GO
     ontology.

> I made a list listGO2GENE to have for each term GO all the genes that belong
> to this term GO. For this, I use ath1121501G02PROBE  or
> ath1121501GO2ALLPROBES. The problem is that the result depends on the choice
> between ath1121501G02PROBE  or ath1121501GO2ALLPROBES.

Since the two environments are different, is this a problem? ;-)

> For example, if I use ath1121501G02PROBE, listGO2GENE["GO:0008104"] gives me
> three genes in the the GO term GO:0008104
> $"GO:0008104"
> [1] "AT1G48090" "AT3G50380" "AT5G24740"
>
> and if I use ath1121501GO2ALLPROBES, I obtain more than 570 genes
> for this GO term...
>
> So what is the right file I have to use to obtain the list that gives me 
> all the genes
> (in TAIR format) for a particular GO term.

It depends a bit on what you want, but I think you want the
*GO2ALLPROBES environment.  This one says, given GO term X, give me
all the probe set IDs annotated at X or at a term that is a descendant
of X.  That is the real annotation list for GO term X.

> Here under, I give you the code  I used  to obtain the list 
> listGO2GENE...may be
> the pb is in my code??But it corresponds to a part of the code GOHyperG, 
> so it
> would be right...
>
> goV<-mget(ls(ath1121501GO2PROBE),ath1121501GO2PROBE)

as.list(someEnv) would be easier here.


> listGO2GENE<- sapply(goV, function(x) {
>   if (length(x) == 0 || is.na(x))
>     return(NA)

Not sure this part does what you want.  The x may have length > 1 in
which case you return NA for x = c(NA, foo, bar, baz), but not for x =
c(foo, bar, NA, baz).  You probably want:
  if (length(x) == 0 || (length(x) == 1 && is.na(x)))

>   lls <- unique(unlist(mget(x, 
> get("ath1121501LOCUSID",mode="environment"))))

If you know the name of the env, there is no reason to use get().

>   lls <- lls[!is.na(lls)]
>   lls
> })

Also, FYI, GOstats has gone through a major revision in the 1.9
release; GOHyperG is deprecated in favor of hyperGTest (see the
vignette).

Also, you didn't tell us versions of anything and that would, in
general, be helpful.

+ seth