[Bioc-devel] R: Re: Deseq2 and differentia expression

Simon Anders anders at embl.de
Sat Jul 12 10:32:43 CEST 2014


Hi Jarod

Mike overlooked one point in your question

On 11/07/14 12:05, jarod_v6 at libero.it wrote:
> If I interested in the genes that have a foldchange more than 0.5 and 2 I need
> to use this comand is it right?

DESeq2 reports al fold changes on a log2 scale. So your limits of 0.5 
and 2 for unlogarithmized fold changes translate to -1 and +1 on the 
log2 scale (because 2^(-1)=0.5 and 2^1 = 1, with '^' meaning 'to the 
power of').

Also, the 'lfcThreshold' parameter wants an _absolute_ log fold change. 
Hence, you want:

ddsNoPrior <- DESeq(ddHTSeq, betaPrior=FALSE)
res <- results(ddsNoPrior, lfcThreshold=1, altHypothesis="greaterAbs")

to get a list of all genes with an absolute log2 fold change greater 
than 1, i.e., all genes with a log2 fold change greater than 1 or less 
than -1, i.e., all genes with fold change below 0.5 or above 2.

Then, in the results table, look at the log2FoldChange column to see 
which genes went up and which went down.

   Simon



More information about the Bioc-devel mailing list