[BioC] Best way of presenting a sub-fraction of edgeR RNAseq gene expression results in a publication

[Gordon K Smyth]

Dear Sindre,

Just run edgeR as usual on all genes.  Run topTags() with n=Inf to get a 
table of DE results for all genes:

   tab <- topTags(yourresults, n=Inf)

Then subset the table for just the 5 genes you are interested in.  For 
example, if your results include a column of gene symbols, this might be:

   tab5 <- tab[tab$table$Symbol %in% mygenes,]
   tab5

When you report the table, there is no need to include the FDR or multiple 
testing column because the genes have been chosen a priori.  Just judge 
significance from the PValue column.

Best wishes
Gordon

> Date: Thu, 30 Jan 2014 03:49:18 -0800 (PST)
> From: "Sindre [guest]" <guest at bioconductor.org>
> To: bioconductor at r-project.org, sindre.lee at studmed.uio.no
> Subject: [BioC] Best way of presenting a sub-fraction of edgeR RNAseq
> 	gene	expression results in a publication
>
>
> Hi!

> We selected a priori 5 genes to analyse and want to include them in a 
> table for publication.
>
> Question is how to present them? Normally we would calculate fold 
> changes (or percent) with standard deviation and then mark if 
> significant.
>
> edgeR, as far as I understand, uses shrunken fold changes and the 
> dispersion is squeezed using an empirical Bayes method. Thus, its hard 
> to manually reproduce the results from logCPM values.
>
> So, whats the best way?
>
> Thank you!
>
> -- output of sessionInfo():
>
> No codes used.
>
> --
> Sent via the guest posting facility at bioconductor.org.

______________________________________________________________________
The information in this email is confidential and intend...{{dropped:4}}