[BioC] DESeq diagnostics

Simon Anders anders at embl.de
Tue Oct 9 22:12:21 CEST 2012


Hi Michael

On 07/10/12 18:17, Michael wrote:
> BTW, is it better to estimate dispersion in paires (treatment1 / control;
> treatment2 / control, ...) or rather for a whole dataset in case of 1
> control / treatment 1-6  (each sample in 2 biol rep).

As you have only two replicates per treatment, you are most likely 
better of if you use your whole data set to get reasonably precise 
dispersion estimates.

However, if there is one treatment where the replicates show stronger 
differences than in the othe treatments, this will increase dispersion 
and hence reduce power for all comparisons. So, better double-check that 
this is not the case, e.g. with the plotPCA function.

   Simon



More information about the Bioconductor mailing list