[BioC] GOseq for enrichment of clustering results

[Alicia Oshlack]

Hi Julie,

In theory you can get your list of "DE" genes any way you like before using
Goseq. The issue however for GOseq is that usually when you perform a
statistical test for DE it is biased towards long and highly expressed
genes. I'm not exactly sure how these features will effect clustering but if
you see a trend when calculating nullp() then it's probably worth using
Goseq to correct your GO testing for biases.

Cheers,
Alicia


On 18/04/12 8:00 PM, "bioconductor-request at r-project.org"
<bioconductor-request at r-project.org> wrote:

> From: <Julie.Leonard at SYNGENTA.COM>
> To: <bioconductor at r-project.org>
> Subject: [BioC] GOseq for enrichment of clustering results
> Message-ID:
> <E6E983882972D447AF74D38898C199631291E2F968 at USETCMSXMB02.NAFTA.SYNGENTA.ORG>
> 
> Content-Type: text/plain
> 
> I am thinking of using GOseq to obtain enriched GO terms for RNA-Seq
> clustering results. I plan to transform the RNA-Seq data before running
> clustering to make it homoskedastic.  When running GOseq, I would provide the
> list of genes that cluster together as the list of "DE genes".
> 
> Question: Is this an appropriate usage of GOseq?  Or can/should I just use a
> standard Fisher's Exact Test?
> 
> Thanks,
> Julie
> 
> Julie Leonard
> Computational Biologist
> Syngenta Biotechnology Inc.
> (919) 281-7449
> julie.leonard at syngenta.com
> 


______________________________________________________________________
This email has been scanned by the Symantec Email Security.cloud service.
For more information please visit http://www.symanteccloud.com