[BioC] DEXSeq - identical p-values for all exons in a gene

[Pillman, Katherine (Health)]

Hi Simon/everyone,

Thanks for your help troubleshooting this, I believe I have figured out the problem.  Embarrassingly, it was a version issue - the version of DEXSeq on my server was not the most current version (it was version 1.7.x).  I should have checked earlier but didn’t think to because I didn't realise there had been a recent version change and my basic analyses ran without problems...  I presume that the way the model needs to be specified changed between the previous (v1.7.x) and current (v1.8.0) versions?

Anyway, I updated BiocGenerics and DEXSeq and now I get unique p-values for each exon in the gene when I specify the models! :)

I noticed that the ‘news’ page for DEXSeq hasn’t been updated yet with the changes for version 1.8.0.  Would you mind please adding the changes?

Thanks again for your help and sorry for the trouble.

Cheers, Katherine

________________________________________
From: Simon Anders [anders at embl.de]
Sent: Monday, 20 January 2014 6:45 PM
To: bioconductor at r-project.org; Pillman, Katherine (Health)
Subject: Re: [BioC] DEXSeq - identical p-values for all exons in a gene

Hi Katherine

On 19/01/14 23:34, Pillman, Katherine (Health) wrote:
> Thanks for your response to my message.  I'm sorry but I don't
> understand what you mean (perhaps this is the root of my problem!).
> You wrote that I did not specify the full model formula to
> estimateDispersions.  I am confused because I thought that was what I
> was doing when I wrote "formula=formulaFullModel" in the line:
>
>>> ecs <- estimateDispersions(ecs, formula=formulaFullModel,
>>> nCores=cores, quiet=FALSE, file='dexseq_progress_report.txt')

Actually, this line is correct and does contain the formula as it
should. Not sure, how I have overlooked this last time. Very sorry!

So, your problem must lie somewhere else. Maybe something is wrong with
the counts. Could you post the output of "countTableForGene" for one of
the genes that have all identical p values?

   countTableForGene( ecs, "A1BG", normalized=FALSE )
   countTableForGene( ecs, "A1BG", normalized=TRUE )

Maybe also post the full code and sample tables once more, because there
is some strange mismatch: The formulas you posted earlier mention a
factor "libType", which does not appear in your sample table. There, you
have, besides "condition", the factors "type" and "end".


  Simon