[BioC] help with DESeq

[Akula, Nirmala (NIH/NIMH) [C]]

Thank you very much Simon for your response.

Nirmala

------------------------------------------------------------------------------------
Nirmala Akula, MS, PhD
Contractor,
Human Genetics Branch
NIMH/NIH
Blg 35, Rm 1A/205
Bethesda, MD - 20892
Phone# 301-451-4258

-----Original Message-----
From: Simon Anders [mailto:anders at embl.de]
Sent: Monday, December 12, 2011 4:26 PM
To: Akula, Nirmala (NIH/NIMH) [C]
Cc: bioconductor at r-project.org
Subject: Re: [BioC] help with DESeq

Dear Nirmala

On 2011-12-12 22:01, Akula, Nirmala (NIH/NIMH) [C] wrote:
>> nisc1Design
>           Condition RIN Age Sex
> bipolar1   Bipolar 8.8  46   1
> bipolar2   Bipolar 6.0  48   1
> bipolar3   Bipolar 8.9  73   1
> bipolar4   Bipolar 8.0  56   2
> control1   Control 8.5  45   1
> control2   Control 8.6  57   1
> control3   Control 8.0  63   1
> control4   Control 8.1  39   2
> control5   Control 7.9  56   2

We only had factorial designs in mind when implementing DESeq's GLM
facility. It may be possible to tweak it to also accept quantitative
covariates but I am not convinced that this would have many applications.

I don't know what exactly you are aiming at in your analysis but hoping
that a covariate has a linear influence on your measured quantity (gene
expression, I presume) seems extremely optimistic in the case of age --
and for RIN, I have no idea why one should expect that.

By the way, if you encode sex with integers, the GLM fitter might
mistake this for a quantitative covariate as well. Better use letters to
be sure that it is treated as a factor.

    Simon