[BioC] GOstats - hyperGTest

[Sean Davis]

On Sat, Mar 22, 2008 at 4:49 PM, Paul Evans <p.evans48 at yahoo.com> wrote:
> Hi,
>
>  I was trying the hyperGTest function and used this simple script:
>
>  ----------------------------------
>  > library("hgu95av2")
>  > library(GOstats)
>  >
>  > chip <- "hgu95av2"
>  > hgCutoff <- 0.2
>  > ont <- "BP"
>  >
>  > xx <- as.list(hgu95av2ENTREZID)
>  > # Remove probe identifiers that do not map to any ENTREZID
>  > xx <- xx[!is.na(xx)]
>  > if(length(xx) > 0){
>  +      # The ENTREZIDs for the first two elements of XX
>  +      xx[1:2]
>  +      # Get the first one
>  +      xx[[1]]
>  + }
>  [1] "652"
>  >
>  > allEntrez <- unlist(xx)
>  >
>  > geneSet <- allEntrez[1:50]
>
>  > params <- new("GOHyperGParams", geneIds = geneSet,
>  +              universeGeneIds = allEntrez, annotation = "hgu95av2", ontology = ont,
>  +              pvalueCutoff = hgCutoff, conditional = FALSE, testDirection = "over")
>  >
>  > hgOver <- hyperGTest(params)
>  Error: value for 'GO:0032650' not found
>  >
>  -------------------------------------------------------------
>
>  I feel that I am making some very elementary mistake, but can't quite put my finger on it. Any hints/documentation that would steer me in the right direction?

Hi, Paul.

I think this is the kind of error you might see when your annotation
package and the GO package are out-of-sync.  You'll need to supply
sessionInfo() output to be sure (and you always should to the list),
but you could also check yourself that the GO and hgu95av2 packages
are from the same bioconductor release.  The easiest way to ensure
that all is well is probably to reinstall them using biocLite().

Sean


Sean