[BioC] subtractmm problem in "affy"

[Ben Bolstad]

Yes you are correct. if you read the 

"affy: built in processing methods"

vignette, it is clearly spelled out in section 6 that  "It is important
to note that not every preprocessing method can be combined together. In
particular the rma backgrounds adjust only PM probe intensities and so
they should only be used in conjunction with the pmonly PM correction."

It would be trivial for someone to modify their local bg.correct.rma to
handle MM probes. Say something like: 

bg.correct.rma <- function(object, ...){
  pm(object) <- apply(pm(object),2,bg.adjust)
  mm(object) <- apply(mm(object),2,bg.adjust)
  return(object)
}
 
or you could use

bgcorrect.method="none"

in the expresso() call


On Thu, 2004-04-22 at 02:40, Anja von Heydebreck wrote:
> I'm wondering whether a problem with Ben's suggestion might be that in
> expresso, the pmcorrect step is applied after background correction (and
> normalization), and the RMA bg correction is applied to the PM
> intensities only. So, as far as I understand, by choosing pmcorrect.method
> other than "pmonly" together with "bgcorrect.method="rma", one would
> subtract not-bgcorrected MM intensities from bgcorrected PM intensities.
> 
> > Date: Wed, 21 Apr 2004 13:55:12 -0700
> > From: Ben Bolstad <bolstad at stat.berkeley.edu>
> > Subject: Re: [BioC] subtractmm problem in "affy"
> > To: "PARK,CHRISTOPHER CHANGSUN" <ccpark at ucla.edu>
> > Cc: bioconductor at stat.math.ethz.ch
> > Message-ID: <1082580912.1635.56.camel at bmbbox.dyndns.org>
> > Content-Type: text/plain
> >
> > the problem is that the medianpolish summarization method takes a log2
> > transformation of the data. if you use subtractmm  there is the
> > possibility of gettting negatives. Perhaps you want to use
> >
> > pmcorrect.method="mas"
> >
> > which will used a modified version of MM. This will not give you the
> > same problem of NA.
> >
> >
> > On Wed, 2004-04-21 at 13:49, PARK,CHRISTOPHER CHANGSUN wrote:
> > > Hi all,
> > >
> > > I am trying to normalize microarray data using the subtractmm method using
> > > the "affy" library.
> > >
> > > The following is my R script:
> > >
> > > set.seed(123)
> > > library(affy)
> > > Data <- ReadAffy()
> > > eset<-expresso(Data,bgcorrect.method="rma",
> > > normalize.method="quantiles",
> > > pmcorrect.method="subtractmm",
> > > summary.method="medianpolish")
> > > write.exprs(eset,file="subtractmm_data.txt")
> > >
> > > During the normalization, I encounter the following error message:
> > >
> > > Error in if (converged) break : missing value where TRUE/FALSE needed
> > > In addition: Warning message:
> > > NaNs produced in: log(x, base)
> > > ...
> > >
> > > The processed data is a grid of NA's.
> > >
> 
> Best,
> Anja
> 
> --
> Dr. Anja von Heydebreck
> Department of Bio- and Chemoinformatics
> Merck KGaA
> 64271 Darmstadt, Germany
> Phone: ++49-6151-723235
> Fax:   ++49-6151-723329
> anja.von.heydebreck at merck.de
> 
> _______________________________________________
> Bioconductor mailing list
> Bioconductor at stat.math.ethz.ch
> https://www.stat.math.ethz.ch/mailman/listinfo/bioconductor
-- 
Ben Bolstad <bolstad at stat.berkeley.edu>
http://www.stat.berkeley.edu/~bolstad