Dear all,
I'd like to use edgeR to find differently expressed genes in my RNAseq
data. My question is which dispersion I should use.
In a forum I read that moderate tagwise dispersion ranks more highly
the DE genes which are consistent in their group - which is what I am
looking for.
How can I achive "moderate" tagwise dispersion or could I use trended
dispersion instead?
My data consists of 18 libraries with 3 treatments (6 libraries each).
In these treatments I have 2 groups of 3 libraries each which are not
exactly replicates but should behave similarly.
I want to test mainly for differences between the groups in each
treatment.
Thanks for your Help!
Kind Regards
Kathrin Ruprecht
Master student
Plant ecological genetics
ETH Zurich
Dear Kathrin,
We almost always recommend tagwise moderated dispersions, but it is
not a
choice between trended and tagwise because the moderation is by
default
towards the trend.
Just follow any of the examples in the User's Guide. Or alternatively
use
the shortcut:
y <- calcNormFactors(y)
y <- estimateDisp(y)
plotBCV(y)
which will doing everything for you.
Best wishes
Gordon
> Date: Mon, 27 Jan 2014 12:23:27 +0100
> From: Ruprecht Kathrin <k.ruprecht at="" gmx.ch="">
> To: <bioconductor at="" stat.math.ethz.ch="">
> Subject: [BioC] which dispersion to choose
>
> Dear all,
>
> I'd like to use edgeR to find differently expressed genes in my
RNAseq data. My question is which dispersion I should use.
> In a forum I read that moderate tagwise dispersion ranks more highly
the DE genes which are consistent in their group - which is what I am
looking for.
> How can I achive "moderate" tagwise dispersion or could I use
trended dispersion instead?
>
> My data consists of 18 libraries with 3 treatments (6 libraries
each). In these treatments I have 2 groups of 3 libraries each which
are not exactly replicates but should behave similarly.
> I want to test mainly for differences between the groups in each
treatment.
>
> Thanks for your Help!
> Kind Regards
>
> Kathrin Ruprecht
> Master student
> Plant ecological genetics
> ETH Zurich
______________________________________________________________________
The information in this email is confidential and
intend...{{dropped:4}}
