[BioC] Which log2FC to report?

Michael Love michaelisaiahlove at gmail.com
Tue Apr 15 17:05:29 CEST 2014


hi Rafael,

On Tue, Apr 15, 2014 at 9:50 AM, Rafael Moreira [guest]
<guest at bioconductor.org> wrote:
>
>
> while in DESeq2, I have:
> rawData <- DESeqDataSetFromMatrix(counts, pd, ~ lane + condition)
> dds <- DESeq(rawData, test='LRT', reduced= ~ lane)
>
> >From the documentation, this seems the right way of getting the DE genes when I want to account for the lane effect. Is this correct? Or would be something like
>
> DESeq(rawData)
>
> sufficient? (and of course getting the results for the proper coefficient)

These are both correct for DESeq2. The first formulation performs the
likelihood ratio test (by default not shrinking LFCs and the second
performs the Wald test by default shrinking LFCs). See ?DESeq for more
information.  Either way results() will give you the log2 fold change
for condition.

Mike



More information about the Bioconductor mailing list