[BioC] Multifactor analysis with deseq

[Simon Anders]

Hi

[Please keep conversations on the list.]

On 25/03/13 19:30, Michael Muratet wrote:
> I thought I would use them in R to manually calculate the GLM where I
> could have access to the model. Or, maybe there is a way to get that
> in deseq2?

What kind of "access" do you need? You can specify a formula (or even a 
model matrix, if you prefer), and you get all coefficients ande 
deviances back.

Of course, you can call glm.fit or statmod::glmnb.fit directly. Then, 
however, you should _not_ use the normalized counts but rather specify 
the log size factors as offsets.

> I tried to install deseq2 and discovered I need R 2.16. CRAN only
> goes up to 15.3, but they have beta r003, is this what you used?

No, I used R 2.15. See the explanations on the Bioc web page on how R 
and Bioc versions are tied together, and how devel packages are handled.

   Simon