[Bioc-sig-seq] using DEXSeq for up and down regulated exons

[Wolfgang Huber]

Dear Jane

thank you for your suggestion on improving the output of DEXSeq, which 
we will take on board.

To fix your immediate problem, I suggest you use the plot functions 
(Section 5 of the vignette) and the 'DEXSeqHTML' in order to create the 
HTML report, from which you can read off the information that you need.

Currently, this is not yet automated, and there is a reason for that: 
speaking of "an exon being up- or down-regulated" only really makes 
sense if a single (or very few) exons change within a transcript of many 
other exons, since that statement in DEXSeq is relative to the average 
of the other exons.

OTOH, if you want to simply know whether exons go up and down, 
irrespective of what the other exons (i.e. "the gene") does, then simply 
use DESeq.

	Best wishes
	Wolfgang




Jul/29/11 11:39 AM, Jane Merlevede scripsit::
> Hello,
>
> I am using DEXseq for differential analysis. I have posted some e-mails
> about it already on this list, but I have more questions !
> My dataset has 2 conditions with 3 biological replicates per condition:
>
> pData(ecs)
>          sizeFactor condition replicate       type
> Raman_1  0.9638822    nonvir         1 paired-end
> HM1_1    1.4000715       vir         1 paired-end
> Raman_2  1.0314049    nonvir         2 paired-end
> Raman_3  1.0734133    nonvir         3 paired-end
> HM1_2    0.7801269       vir         2 paired-end
> HM1_3    0.9488409       vir         3 paired-end
>
>
> I found the number of differentially expressed exons. I would like to know
> also which are down and which are up regulated. This information is given by
> foldChange, which is not provided in DEXSeq (res1).
>
> res1<- DEUresultTable(ecs)
> head(res1)
>                     geneID exonID dispersion_CR_est dispersion    pvalue
> EHI_000010:001 EHI_000010   E001                NA 0.02377712        NA
> EHI_000130:001 EHI_000130   E001        0.07466945 0.07466945 0.4867354
> EHI_000130:002 EHI_000130   E002        0.12897281 0.12897281 0.9427953
> EHI_000130:003 EHI_000130   E003        0.01960222 0.02031871 0.8432791
> EHI_000130:004 EHI_000130   E004        0.06733720 0.06733720 0.7042977
> EHI_000240:001 EHI_000240   E001        0.04375400 0.04375400 0.7493987
>                   padjust
> EHI_000010:001        NA
> EHI_000130:001 0.8639624
> EHI_000130:002 0.9927663
> EHI_000130:003 0.9756197
> EHI_000130:004 0.9503086
> EHI_000240:001 0.9611397
>
> So I would like to know, if it could be possible to have baseMean,
> baseMeanA, baseMeanB, foldChange, log2FoldCHange, resVarA and resVarB as in
> DESeq?
>
> Since this does not seem to be implemented in DEXSeq, I should start with
> the read counts per exon table to get it, but I can't find it neither. I
> think it is use in the estimateSizeFactors function, but I don't manage to
> access it. It's a pity, I will have to reconstruct the table...
>
>
> Thank you for your help,
> Jane Merlevède
>
> 	[[alternative HTML version deleted]]
>
>
>
>
> _______________________________________________
> Bioc-sig-sequencing mailing list
> Bioc-sig-sequencing at r-project.org
> https://stat.ethz.ch/mailman/listinfo/bioc-sig-sequencing


-- 


Wolfgang Huber
EMBL
http://www.embl.de/research/units/genome_biology/huber