[BioC] DESeq for differentiating highly expressed genes

[Steve Lianoglou]

Hi,

On Tue, Feb 11, 2014 at 6:25 AM, Liang, Chun <liangc at miamioh.edu> wrote:
> When DESeq is used for our RNA-Seq data analysis of two conditions, each of
> them has three biological replicates, we got a result better than Cuffdiff2
> and CLCBio Genomics Workbench, verified by RT-qPCR.
>
> Based on the normalized counts and adj-p values, can we also tell with
> confidence which genes are highly expressed and which genes are rarely
> expressed within the same conditions.  I knew that DESeq have library size
> factor in normalization, but do not have gene length factor in
> normalization like Cuffdiff2.  Any comment will be highly appreciated.

You could always divide the normalized counts by the K in RPKM do get
a similar number, no?

For instance, edgeR has an `rpkm` function to do the same -- you can
take inspiration from the code there to see how to do it.

-steve

-- 
Steve Lianoglou
Computational Biologist
Genentech