[BioC] passing my data to siggenes package
Holger Schwender
holger.schw at gmx.de
Thu Oct 27 13:33:39 CEST 2005
Hi Madhurima,
to your first email: data in sam(...) must not necessarily be a matrix or
data frame. It can also be an exprSet object. If data is specified by an
exprSet object and if you have already specified the class labels in the
corresponding phenoData object, you can also specify cl by the name of
column of the phenoData matrix that contains the class labels.
to the fudge factor problem: This problem usually occurs when you have a too
small number of genes (please recall that you actually should have a few
hundred genes when applying SAM). In the computation of the fudge factor as
described in the Tusher paper, the standard deviations are divided into
intervals (usually, 101 intervals, fudge2 also allows a much smaller number
of intervals, namely down to 5, where in each of these intervals have to be
5 values) to compute the fudge factor. If there are less than 25 genes or
more precisely less than 25 genes with differing standard deviations, fudge2
stops and returns the error you have got. You can solve this problem by
either using more genes (e.g., don't filter before SAM) or by setting
s.alpha in sam(...) to a reasonable value. Tibshirani, e.g., uses in the
Excel implementation s.alpha=0.5.
Best,
Holger
> --- Ursprüngliche Nachricht ---
> Von: Sean Davis <sdavis2 at mail.nih.gov>
> An: madhurima bhattacharjee <madhurima_b at persistent.co.in>
> Kopie: Bioconductor <bioconductor at stat.math.ethz.ch>
> Betreff: Re: [BioC] passing my data to siggenes package
> Datum: Wed, 26 Oct 2005 07:43:58 -0400
>
> On 10/26/05 6:53 AM, "madhurima bhattacharjee"
> <madhurima_b at persistent.co.in> wrote:
>
> > Hello Sean,
> > Thanks for the response.
> > Given below is the part of the code and the error that I am getting:
> >
> >> sam.out <- sam(madhu1,my.cl)
> > Error in fudge2(r, s, alpha = s.alpha, include.zero = include.zero) :
> > For the computation of the fugde factor,
> > there should be at least 25 genes with differing standard deviations.
> >
> > Here madhu is the matrix and my.cl is the class vector.
> > What is the problem with the data?
> > Could you please tell me how to solve this.
>
> I don't use siggenes often enough to know if this message means more than
> it
> says, so I can't be sure. But, what does madhu look like? How did you
> make
> that matrix? The error message is implying that there are very few genes
> that show any variation. Perhaps someone who uses siggenes more often can
> be more specific.
>
> Sean
>
> _______________________________________________
> Bioconductor mailing list
> Bioconductor at stat.math.ethz.ch
> https://stat.ethz.ch/mailman/listinfo/bioconductor
>
--
More information about the Bioconductor
mailing list