[R] samr-package: problem with large sample size multiclass data

Rainer Tischler rainer_t62 at yahoo.de
Tue Feb 24 18:53:05 CET 2009


Dear all,

I'm using the samr-package to identify significantly differentially expressed genes in microarray data.
So far, I had no problems, but when I used a large multiclass data set with 327 samples, I obtained the following error/warning message:

Warning message:
Inf
In factorial(length(y)) : value out of range in 'gammafn'

Since y is the label vector and has length 327, the factorial is indeed a very large number. Am I doing something wrong or is the samr-package just not appropriate for large sample size data sets?
Though I only obtain a warning message and no error message, the output of the algorithm is obviously incorrect, because it tells me that the data set does not even contain a single significantly differentially expressed gene (other methods detect hundreds of them on this data set, and when reducing the number of samples even samr detects many differentially expressed genes).
I'm using R 2.8.0 on Windows XP. The response-type in samr was set to "Multiclass".

Many thanks,
Rainer

WikipediaWictionaryChambers (UK)Google imagesGoogle defineThe Free DictionaryJoin exampleWordNetGoogleUrban DictionaryAnswers.comrhymezone.comMerriam-Webster<>0
wvcidfjoguarm






More information about the R-help mailing list