[BioC] Importing RMA processed data into limma for eBayes

[Thornton, Matthew]

Dear List Patrons,

I am processing some microarray data obtained with Affymetrix Mouse Gene 2.0 ST chips. Using the xps package, I have performed step-wise RMA processing - Background correction, Quantiles Normalization, and Median Polish summarization. The output of xps is a tab-delimited text file with 33,960 non-log base 2 transformed expression measures, the UnitName (an 8 digit number starting at 17550607) and a Unit ID (starts at 0 but isn't continuous).  What I would like to do is import this data into the limma package, from my output text file.  I know that I can make an expressionset and use this expressionset directly after making a design matrix, via the instructions here:

http://permalink.gmane.org/gmane.science.biology.informatics.conductor/50591

I would like to add the gene names into the expressionset so that I can make volcano plots and graphs with limma, in accord with the limma users guide.

I know how to access the gene names and relate them to the UnitName code copied from here: (http://biowhat.ucsd.edu/homer/basicTutorial/affymetrix.html)

library(mogene20sttranscriptcluster.db)
library(pd.mogene.2.0.st)

mat1 <- as.matrix(exprs_mp)
post_rma  <- ExpressionSet(assayData=mat1, annotation="mogene.2.0.st")

my_frame <- data.frame(exprs(post_rma))

Annot <- data.frame(ACCNUM=sapply(contents(mogene20sttranscriptclusterACCNUM), paste, collapse=", "), SYMBOL=sapply(contents(mogene20sttranscriptclusterSYMBOL), paste, collapse=", "), DESC=sapply(contents(mogene20sttranscriptclusterGENENAME), paste, collapse=", "))

all <- merge(Annot, my_frame, by.x=0, by.y=0, all=T)

write.table(all,file="21Dec13_data.ann.txt",sep="\t")

Is there a way to modify the above code so that I can add the gene names and gene information to my expresionset, in a way that makes them AS accessible to the limma package as they would be if I had used the affy package to process the data initially?

Also, I would like to compare the output of VSN with RMA using eBayes in limma. I don't wish to import the raw intensities into limma, not because I couldn't use the same methods (vsn is in limma), only because I would have to start my data processing over. Any help or comments is greatly appreciated.

Thank you.

Sincerely,

Matt


Matthew E. Thornton

Research Lab Specialist
Saban Research Institute

USC/Children’s Hospital Los Angeles
513X,  Mail Stop 35
4661 W. Sunset Blvd.
Los Angeles, CA 90027-6020

matthew.thornton at med.usc.edu