[BioC] hyperGTest results I do not understand

Seth Falcon sfalcon at fhcrc.org
Tue Feb 27 20:42:39 CET 2007


ariel at df.uba.ar writes:

> Hi Seth, and thanks for your prompt response
>
> I do not think that this is a conditional analysis issue because
> the node I am focus on has no children.
>
> I did forget to take into account a specific ontology to make
> the contingency table. However I am still puzzled because even if I do
> filter on Ontology, the expected count value inferred from the new table
> does not agree with the one reported by GOstats either.
>
> New contingency table for BP:
>
>          selected  ~selected
> gonode   13           2
> ~gonode 1230       1334
>
> which gives ExpCount=7.229 (and not 13.8)
>
>
> So, if you do not mind, I accept your offer and I will send you, in
> private email, my selected and universe lists for you to check them.
> Thank you so much in advance

There was a bug that may be related to this issue that has been
fixed.  The current versions for use with R 2.4.x are GOstats 2.0.4
and Category 2.0.3.

> I am running R 2.4 on Fedora 5 and here is the output of sessionInfo():
>
> other attached packages:
> RColorBrewer      GOstats     Category         KEGG         RBGL           GO
>      "0.2-3"      "2.0.0"      "2.0.0"     "1.14.0"     "1.10.0"     "1.14.0"

The fine print of my offer to look at your data was that you confirm
the issue with the latest versions ;-)

Can you try getting the latest and giving this one more try.  You
could update like this:

    source("http://bioconductor.org/biocLite.R")
    biocLite(c("Category", "GOstats"))

Or to get updates for all of your installed packages (may take awhile,
but is a good idea since updates usually mean bug fixes):

    library("Biobase")
    update.packages(repos=biocReposList())


Best,

+ seth



More information about the Bioconductor mailing list