[BioC] GO analysis

[Sean Davis]

Jose,

Assuming that you have an annotation data package built for your data and
the GO data package (from Bioconductor), you can use a combination of
lookups from GO to get terms of interest and then use lookup of those terms
to see if any of your probes are annotated with them (see ?lookUp and
?getGO and the package annotate).  Doing the latter on subsets of your data
and the whole data set should allow you to form the EASE score (or fisher
exact test, etc.). However, know that the annotation packages use locuslink
information and the other programs may use other (with respect to GO) and
perhaps more complete information and that the answers that you get from the
other programs will likely be slightly different.

Sean
-- 
Sean Davis, M.D., Ph.D.

Clinical Fellow
National Institutes of Health
National Cancer Institute
National Human Genome Research Institute

Clinical Fellow, Johns Hopkins
Department of Pediatric Oncology
-- 



On 12/12/03 12:36 PM, "Jose Duarte" <jose.duarte at human-anatomy.oxford.ac.uk>
wrote:

> Hi
> 
> I would like to do some post annotation analysis on my gene list from an
> Affy experiment. Basically we want to see if the set of genes has any
> statistically over represented GO ids in comparison to the total of the
> genes in the chip. An over representation probability should be given to
> each of the categories.
> 
> This is in line with what some packages already do: EASE, GeneMerge,
> GOminer, fatiGO (web)... But I was wondering if somebody did something
> like this already in Bioconductor or maybe it is not a good idea to
> implement this in R as it is not the fastest text searcher.
> 
> Any suggestions welcomed
> 
> 
> Thanks!
> 
> Jose
> 
> 
> 
> 
>