[BioC] combineAffyBatch problem

Gabor Halasz gh74 at columbia.edu
Fri Mar 9 07:22:02 CET 2007


James W. MacDonald <jmacdon at ...> writes:

> Catherine Dumur wrote:
> > I understand that every time I run the function, a new .CDF object is 
> > created and this takes a considerable amount of time. I was wondering if 
> > we can re-use an existing "newcdf" object when performing new analysis 
> > where the same two types of arrays are used. In other words, since I am 
> > always combining the same 2 types Affy chips, the newcdf object should 
> > be the same. If I use that object from previous sessions, I shouldn't be 
> > creating a new one every time.
> 
> You are correct, you _shouldn't_ be creating a new one every time. 
> Instead, you should save() the newcdf object and load() it back in when 
> you need it. An example would be:
> 
> save(newcdf, file="newcdf.Rda")
> load("newcdf.Rda")


Dear James & Bioconductor group,

I'd like to ask a follow-up on the reply you posted here.  You suggested 
saving and re-loading the new cdfenv environment created by 
combineAffyBatch, but i can't figure out how to apply the new cdfenv to
additional AffyBatch objects. In particular, this is wrong:

-------------------------------------------
library(affy)
library(limma)
library(hgu95acdf) 
library(hgu95av2cdf)
library(matchprobes)
library(hgu95aprobe)
library(hgu95av2probe)

targets <- c("GSM44138.CEL", "GSM44275.CEL")
affyObject1 <- ReadAffy(filenames=targets[1])
affyObject2 <- ReadAffy(filenames=targets[2])

res <- combineAffyBatch(list(affyObject1, affyObject2),
probepkg=c("hgu95aprobe", "hgu95av2probe"), newcdf="hgu95a.hgu95av2")

hgu95a.hgu95av2 <- res$cdf

eset.combcdf1 <- rma(res$dat)   ###correct
eset.combcdf2 <- justRMA(filenames=targets, cdfname="hgu95a.hgu95av2")
#NOT correct
-------------------------------------------

The last line executes without complaint, but the expression values of 
eset.combcdf2 are completely wrong (i.e. completely uncorrelated with 
the results of simply doing rma() on the CEL files individually). 

So my question: why is the last line in the code wrong, and how do i
correctly reuse the combined cdf environment?

Thank you!  

Gabor



More information about the Bioconductor mailing list