[BioC] ANOVA test after VST in DESeq

Simon Anders anders at embl.de
Thu Mar 14 18:57:33 CET 2013


Hi

On 14/03/13 16:57, Liu, XiaoChuan wrote:
> Here are my code for ANOVA test after VST in DESeq,

 > Do you think it is right to do ANOVA test by VST? Thanks!

Doing a VST followed by ANOVA is not entirely wrong but it is certainly 
less suitable then a GLM. I don't see why you don't want to do your 
analysis in the manner suggested in the vignette.

You ANOVA code gives you four p values per gene, and, of course, you can 
get four equivalent p values with GLM tests, too. The reason why I 
haven't told you how is that I doubt that you are interested in all four 
of them, and as you have not told us what biological question you want 
to address, I cannot say which one would be the one relevant for you. 
So, I stick to my previous advice: Please read up on the concept of 
"interactions" in linear models and ANOVA, because without that, it is 
unlikely that you manage to make sense of your data. And once you know 
whether you are looking for an interaction or a main effect, ask again.

   Simon



More information about the Bioconductor mailing list