[BioC] (GC)RMA estimates without summerization?
Ben Bolstad
bmb at bmbolstad.com
Tue Feb 12 03:51:47 CET 2008
Answering your second question first: In the case of the GCRMA
background correction all preprocess() is doing is calling
bg.adjust.gcrma via a wrapper bg.correct.gcrma() with the default
arguments. Unfortunately I don't think I provided a method for passing
GCRMA arguments via preprocess() but a quick and dirty thing to do in
your current session would be to do something like this
bg.correct.gcrma <- function(object){
bg.adjust.gcrma(object, fast=FALSE)
}
then call preprocess().
I'm not sure what you mean in the second case by mixing log and non log
scale values. I don't see that happening when I run preprocess(). In any
case, perhaps you are really interested in only the PM or MM probes, in
which case you should be using the pm() or mm() accessors to just get
those. In that case you could then use probeNames() to get the
corresponding names for each probe. Note that exprs() as you are using
it would get pm, mm and control probes (some of which may not have been
adjusted by the pre-processing methods you selected).
Best,
Ben
Mini-Rant:
Note that I am not a big fan of your usage of the terminology
"expression estimates per probe". Really in my view summarization is a
vital part of generating an expression value, instead really what you
are asking for is preprocessed probe intensities.
On Mon, 2008-02-11 at 21:41 +0100, Hooiveld, Guido wrote:
> Thanks Ben, got it working fine for RMA!
>
> Two more questions, though (I hope you don't mind):
> - Output of preprocess is an AffyBatch object; how to save the
> expression estimates?
> I tried
> > esetrma <- preprocess(rawdata)
> > write.table(exprs(esetrma), file="RMA_no_summary.tx", sep="\t")
> but that produces a huge file, and when viewed it contains both
> 'normal'and log2 values; the log2-values are obviously the expression
> estimates, but what are the 'normal' values? These are likely not
> needed. And how to put the probe ID in the first column, i.e. how to
> list the expression estimates per probe?
>
>
> - I would like to apply the emperical Bayes estimate for bakcground
> correction of GCRMA in stead of the miximum likelyhood estimate (MLE).
> Normally you would do this by specifying the argument "FAST=false":
> esetgcrma <- gcrma(rawdata, fast=FALSE)
> How to do this in preprocess() [or threestep()]? I assume both functions
> apply the default settings of gcrma, which is fast=TRUE.
> This doesn't work:
> > esetgcrma <- preprocess(rawdata, background.method = "GCRMA",
> fast=FALSE)
> Error in preprocess(rawdata, background.method = "GCRMA", fast = FALSE)
> :
> unused argument(s) (fast = FALSE)
>
> Thanks again!
> Guido
>
>
>
> > -----Original Message-----
> > From: bmb at bmbolstad.com [mailto:bmb at bmbolstad.com]
> > Sent: 11 February 2008 19:56
> > To: Hooiveld, Guido
> > Subject: Re: [BioC] (GC)RMA estimates without summerization?
> >
> > Try preprocess() in affyPLM, it should allow you to do
> > anything that you can do with threestep() except the
> > summarization step.
> >
> > Best,
> >
> > Ben
> >
> >
> >
> >
> > > Dear list,
> > >
> > > I would like to obtain the GCRMA and RMA expression estimates per
> > > individual probe (thus not per probe set). In other words, I would
> > > like to apply the [default] (GC)RMA background correction and
> > > normalization, but not the summerization step (median
> > polish). How to do this??
> > > I searched of course the mail archive but found no answer on this.
> > >
> > >
> > > I tried the 'expresso' and 'threestep' functions of
> > AffyPLM, but for
> > > summerization it is not explicity stated that
> > "summerization" argument
> > > can be "FALSE" or "NO" (or something else).
> > >
> > > When I nevertheless try to execute threestep I get the
> > following error:
> > >> esetrma <- threestep(rawdata, background.method = "RMA.2",
> > > normalize.method="quantile", summary.method=FALSE) Error in
> > > get.summary.code(summary.method) :
> > > FALSE is not a valid summary method. Please use one of:
> > > median.polish tukey.biweight average.log rlm log.average log.median
> > >
> > > ...clearly indicating that threestep the "summary method"
> > can not be
> > > omitted (as is also stated in the vignette).
> > >
> > >
> > >
> > >
> > >
> > > When I try to do this:
> > >> esetrma <- expresso(rawdata, bgcorrect.method = "rma",
> > > normalize.method="quantiles", pmcorrect.method="pmonly",
> > > summary.method=FALSE)
> > > background correction: rma
> > > normalization: quantiles
> > > PM/MM correction : pmonly
> > > expression values: FALSE
> > > background correcting...done.
> > > normalizing...done.
> > > Error in function (classes, fdef, mtable) :
> > > unable to find an inherited method for function "computeExprSet",
> > > for signature "AffyBatch", "character", "logical"
> > >
> > > ... I got an error likely NOT related to the summary.method option,
> > > but to something else?? But I expect then even when solving this
> > > problem 'expresso' won't allow skipping the summerization step.
> > >
> > >
> > > Thus, how to omit "median polish" in RMA and GCRMA?
> > >
> > >
> > > TIA,
> > > Guido
> > >
> > >
> > >
> > >> sessionInfo()
> > > R version 2.6.0 (2007-10-03)
> > > i386-pc-mingw32
> > >
> > > locale:
> > > LC_COLLATE=English_United Kingdom.1252;LC_CTYPE=English_United
> > > Kingdom.1252;LC_MONETARY=English_United
> > > Kingdom.1252;LC_NUMERIC=C;LC_TIME=English_United Kingdom.1252
> > >
> > > attached base packages:
> > > [1] splines tools stats graphics grDevices utils
> > datasets
> > > methods base
> > >
> > > other attached packages:
> > > [1] medicagocdf_2.0.0 affyPLM_1.14.0 gcrma_2.10.0
> > > matchprobes_1.10.0 affy_1.16.0
> > > [6] preprocessCore_1.0.0 affyio_1.6.1 Biobase_1.16.0
> > >
> > >
> > >
> > > ------------------------------------------------
> > > Guido Hooiveld, PhD
> > > Nutrition, Metabolism & Genomics Group Division of Human Nutrition
> > > Wageningen University Biotechnion, Bomenweg 2
> > > NL-6703 HD Wageningen
> > > the Netherlands
> > > tel: (+)31 317 485788
> > > fax: (+)31 317 483342
> > > internet: http://nutrigene.4t.com <http://nutrigene.4t.com/>
> > > email: guido.hooiveld at wur.nl
> > >
> > >
> > >
> > > [[alternative HTML version deleted]]
> > >
> > > _______________________________________________
> > > Bioconductor mailing list
> > > Bioconductor at stat.math.ethz.ch
> > > https://stat.ethz.ch/mailman/listinfo/bioconductor
> > > Search the archives:
> > > http://news.gmane.org/gmane.science.biology.informatics.conductor
> > >
> >
> >
> >
> >
>
> _______________________________________________
> Bioconductor mailing list
> Bioconductor at stat.math.ethz.ch
> https://stat.ethz.ch/mailman/listinfo/bioconductor
> Search the archives: http://news.gmane.org/gmane.science.biology.informatics.conductor
More information about the Bioconductor
mailing list