[BioC] Error in PLW package plw()

davidl at unr.nevada.edu davidl at unr.nevada.edu
Thu Mar 26 14:47:38 CET 2009


Thank you Martin,

    It's fixed in the new version, and I'll be sure to use sessionInfo() in the
future.  I'm also interested in what everyone thinks about performing
statistical tests on probe level data and then summarizing as compared to
summarizing probe level data and then performing statistical tests for
differential expression.  It seems like PLW is more accurate, but the
interpretation is kind of weird because I'm mixing biological and technical
variation in one test.  Thank you again,

Dave

Quoting Martin Morgan <mtmorgan at fhcrc.org>:

> Hi Dave --
>
> I think your plw is out of date; on this page
>
> http://bioconductor.org/packages/2.3/bioc/html/plw.html
>
> plw is at 1.2.0. Try the instructions here
>
> http://bioconductor.org/docs/install/
>
> to 'Update installed Bioconductor packages'
>
> and...
>
> davidl at unr.nevada.edu writes:
>
> > Hello list,
> >
> >     I am giving "Probe Level locally moderated median-t" (PLW) a try
> > because the ROC curves on the spike-ins looked promising on the paper
> > introducing it
> > (http://www.pubmedcentral.nih.gov/articlerender.fcgi?artid=2358895).
> > First, I think that the function plw() is using a background
> > subtraction function that has been removed from the affy package since
> > plw() is giving me this message:
> >
> > plwFit<-plw(rawdata, design=design, contrast=contrast, epsilon=1e-05)
> > Background correcting and normalizing.  Error: could not find function
> > "bg.correct.rma2" Error in pm(bg.correct.rma2(x, bgtype = 2)) : error
> > in evaluating the argument 'object' in selecting a method for function
> > pm'
> >
> > Is there anything I could do about this (an easy way to make it call
> > bg.correct.rma() instead)?  This is in R 2.8.1, plw_0.99.2, and affy
> > 1.20.2.  Second, I currently have a dataset consisting of 4 affymetrix
>
> ... paste the output of sessionInfo(), it'll be more comprehensive,
> accurate, and easy.
>
> Martin
>
> > mouse 4302 arrays, but 3 are controls and 1 is from a mutant.  Plans
> > are to add 2 more mutant arrays later, but preliminary data is being
> > requested for now.  I was wondering if anyone could weigh in on what
> > they might consider the most accurate ([True negatives+True
> > Positives]/total) method for deriving measures of differential
> > expression on this platform given the lack of replicates in the second
> > group.  What do you think about PLW (I'm not a seasoned statistician,
> > so I probably missed some things when reading the PLW paper)?  Thank
> > you very much in advance for any help given.  I really appreciate all
> > the help I've received from this list.  Cheers,
> >
> > Dave
> >
> > _______________________________________________ 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
>
> --
> Martin Morgan
> Computational Biology / Fred Hutchinson Cancer Research Center
> 1100 Fairview Ave. N.
> PO Box 19024 Seattle, WA 98109
>
> Location: Arnold Building M2 B169
> Phone: (206) 667-2793
>



More information about the Bioconductor mailing list