An embedded and charset-unspecified text was scrubbed... Name: not available Url: https://stat.ethz.ch/pipermail/bioconductor/attachments/20060220/ 3fd023e3/attachment.pl

Pedro, You can use the the rma background correction and normalization routines on a matrix. bg.adjust() implements basically the same background correction functionality that is carried out by the rma() command (except that rma() is implemented in compiled code). The easiest way for you to do this would be my.matrix.bg.rma <- apply(my.matrix, 2, bg.adjust) my.matrix.norm <- normalize.quantiles(my.matrix.bg.rma) Ben On Mon, 2006-02-20 at 15:23 +0100, Pedro L?pez Romero wrote: > Hi all, > > I want to normalize some data comming from a single chanel non affy > platform.- Id like to use the rma( ) function, but I would need to create > first an Affybatch object from the data that I have, a task that I find a > bit difficult to do given the data that I have. I think that would be much > more easier to use the bg.adjust ( ) internal function, but Im not sure if > I can apply this function directly to my data (a matrix with just raw > foreground intensities). > > Ive read the help file of this function but I still have some doubts. I > have understood that bg.adjust ( ) is the internal function that uses rma > ( ) to correct for the background bias. I figure that the input that it > needs is just a matrix with the raw intensities in natural scale and from > here, the function estimates the signal intensities corrected by the > background without using the background measures at all. Am I right?. My > data does not have PM and MM intensities, all the values are PM and I > have a probe per gene, Would still it be correct under theses cirmcumstances > using the bg.adjust ( ) function? > > > Other authors suggest to estimate the true signal intensities just by > substracting the median of the background probe intensities from the mean of > the foreground ones, and setting the negative values, lets say 0.5. I think > that this method is not completely correct, whats your opinion about > this?.- > > Moreover, I want to use the quantiles method to normalize the same data, > would it be right to apply the normalize.quantiles ( ) function directly to > the previous background corrected matrix?- > > Thanks a lot.- > > Pedro > > > > > > [[alternative HTML version deleted]] > > _______________________________________________ > Bioconductor mailing list > Bioconductor at stat.math.ethz.ch > https://stat.ethz.ch/mailman/listinfo/bioconductor

Pedro L?pez Romero <plopez at="" cnic.es=""> writes: Pedro, > > I want to normalize some data comming from a single chanel non affy > platform.- Id like to use the rma( ) function, but I would need to create > first an Affybatch object from the data that I have, a task that I find a > bit difficult to do given the data that I have. I think that would be much > more easier to use the bg.adjust ( ) internal function, but I m not sure if > I can apply this function directly to my data (a matrix with just raw > foreground intensities). > you can not use RMA for your data, since it is designed for Affymetrix arrays, where you always multiple probes per gene. You should have a look at the users guide of the package limma and at the man pages of the functions described therein. You will find several methods for background correction and normalisation. For single channel data, you should have a look at chapter 9 of the limma users guide. Best, Georg