[BioC] rma on new samples

Kuhn, Max Max.Kuhn at pfizer.com
Thu Mar 31 19:13:33 CEST 2005


All,

I'm using R v2.0.1, affy v1.5.8 and Biobase v1.5.0. 

I'm using Affy data for classification. I have a training data set which I
use to process the data (rma, center, scale etc.) for use with various
models. Once I get the models trained, I'll be applying them to future data
as they come in.

During the data processing, I can save the information about the training
set - for example, the means and variances for each probe. I can use these
to center and scale new samples without problems. 

I'm not quite sure to do about the rma step though. The quantile
normalization was not hard to adapt based on modifying limma's
normalizeQuantiles function. In that case, I can just feed in a
pre-normalized sample, get its quantiles and normalize the new samples to
this reference distribution. The other steps in rma do not appear to be so
simple, such as the median polish.

Has anyone implemented code to apply rma on new samples based on previous
data? If not, any suggestions on the approach?

Thanks in advance,

Max


LEGAL NOTICE\ Unless expressly stated otherwise, this messag...{{dropped}}



More information about the Bioconductor mailing list