[BioC] Multifactor analysis with deseq

[Simon Anders]

On 22/03/13 21:06, Simon Anders wrote:
>> Does the formula in the fitNbinomGLMs() function work the same way as
>> lm(), e.g., do I have to explicitly include interactions or will it
>> calculate them for me?
>
> Yes, both fitNbinomGLMs and lm internally call model.matrix, which, as
> you know, adds an intercept automatically

Typed too quickly. No, interactions have to be added explicitly (with 
'*' instead of '+'). To test for interaction effects, compare a full 
model with main effects and interaction (count ~ culture + genotype * 
day ) against one with only main effects (count ~ culture + genotype + 
day ).

   Simon