[BioC] hyperG on custermized categories

Seth Falcon sfalcon at fhcrc.org
Fri Dec 22 19:23:53 CET 2006

Hi Weiwei,

"Weiwei Shi" <helprhelp at gmail.com> writes:
> Instead of using GO or KEGG categories, is it possible to use
> customerized categories. For example:
> CC1 : gene1, gene2, gene3
> CC2 : gene4, gene5..
> instead of
> GO1: gene1, gene2

The hyperGTest code in the Category (and more specialized version in
the GOstats) package only works with "predefined" categories.
Currently, the code supports GO, KEGG, and PFAM.  

However, the code was designed to be straight forward to extend to add
new categories.  If you read the source code implementing KEGG and
PFAM it should be fairly clear what is needed to add support for a new

> For this purpose, I made a code from basic Hypergeometric {stats} by
> using dhyper. But what I got is probability from dhyper. However, I
> really need p-value instead of probability. Should I use phyper
> instead?

You can also look at how the hyperGTest method is implemented in the
Category package.

+ seth

More information about the Bioconductor mailing list