[BioC] P-values returned from siggenes's sam() function

[Holger Schwender]

Hi Michal,

it is the q-value as defined by John Storey. And the values of q-values will be the same as the ones from John's package qvalue if there are no tied raw p-values. If there are ties, then the values might slightly differ because of a slightly different handling of tied p-values.

Best,
Holger


-------- Original-Nachricht --------
> Datum: Mon, 4 May 2009 12:41:02 -0400
> Von: Michal Blazejczyk <michal.blazejczyk at mail.mcgill.ca>
> An: Holger Schwender <holger.schw at gmx.de>
> CC: bioconductor at stat.math.ethz.ch
> Betreff: Re: P-values returned from siggenes\'s sam() function

> Hi Holger,
> 
> Thank you for the answer!
> 
> I've got one more small question: is the FDR algorithm used in SAM
> like one of the commonly known DFS's (Bonferroni, Benjamini-Hochberg,
> Sidak), or is it a special FDR developed for SAM?
> 
> Best regards,
> Michal Blazejczyk
> 
> 
> 
> Holger Schwender <holger.schw at gmx.de> wrote:
> > If you are okay with the assumption that all genes follow the same null
> > distribution (and I guess so, as you are using sam), then yes.
> 
> > In fact, these raw p-values are used to compute the q-values
> > (result at q.value), i.e. FDR adjusted p-values.
> 
> > Holger
> 
> > -------- Original-Nachricht --------
> >> Datum: Thu, 30 Apr 2009 13:45:25 -0400
> >> Von: Michal Blazejczyk <michal.blazejczyk at mail.mcgill.ca>
> >> An: bioconductor at stat.math.ethz.ch
> >> CC: Holger Schwender <holger.schw at gmx.de>
> >> Betreff: P-values returned from siggenes\'s sam() function
> 
> >> Hi everyone,
> >> 
> >> I have a fairly technical question.  We are calling the sam() function
> >> from packages siggenes:
> >>   result <- sam( ...... )
> >> and then getting the p-values like this:
> >>   result at p.value
> >> 
> >> The Help for function sam() says that these are "raw, unadjusted
> p-values
> >> of the genes".
> >> 
> >> Does it mean an FDR function such as mt.rawp2adjp() from multtest can
> be
> >> safely used on these values?
> >> 
> >> Best,
> >> Michal Blazejczyk
> >> FlexArray Lead Developer
> 

--