[BioC] tilingArray normalizeByReference Data Scaling

[Wolfgang Huber]

May/15/12 4:00 AM, Dario Strbenac scripsit::
> Thanks for the clarification. Yes, I was confused which one of the options in
> Figure 5 of the article were implemented in the tilingArray packages.
>
> What I meant by the difference in log bases is :
>
> refSig = rowMeans(log2(exprs(reference)[pm, , drop = FALSE]))
> whereas
> ybg[, j] = tapply(log(exprs(x)[background, j], 2), strata, genefilter::shorth,
> tie.action = "min")
>
> ?log says the default base is e. So is there a mathematical reason to use base
> 2 for reference probes, and base e for background probes ?

Hi Dario

I see. The second argument of 'log' is 'base', which in the ybg<- line 
above is set to 2. In fact base=2 is used in all calls to the 'log' 
function in the package.

Best wishes
	Wolfgang

Wolfgang Huber
EMBL
http://www.embl.de/research/units/genome_biology/huber