[BioC] How to filter genes according to the intensities of both channels for cDNA array?

[Sean Davis]

On May 6, 2005, at 2:46 AM, Xiao Shi wrote:

> Hi everybody,
> I am working on 10 GPR files from a time course cDNA array
> experiment.Aftersome normalization steps,i got the M and A for 10
> arrays.And then i want to apply the quality control procedures to flag
> logratio(M) for those spots to Na whose inteinsities form either  
> channel is
> lower than 200.what can i do?
>
> The procedures i did are as follows:
> library(marray)
> file=c('2000009478.gpr','2000009471.gpr','2000009472.gpr','2000009473.g 
> pr','
> 2000009479.gpr','2000009466.gpr','2000009467.gpr','2000009468.gpr','
> 2000009480.gpr','2000009481.gpr')
> target<-read.marrayInfo(fname="e:/gpr/zpz/zpz-target.txt")
> ##Read probe related information
> galinfo<-read.Galfile("GAL-13K.txt",path="e:/gpr/zpz")
> ##Read expression data
> zraw=read.GenePix
> (fnames=file,path="e:/gpr/zpz/ 
> results",layout=galinfo$layout,gnames=galinfo$gnemes,target=target)
>
> ##normalization on print tips
> zraw.norm=maNormMain(zraw)
>
> ##Between slides normalization
> zraw.normg=maNormScale(zraw.norm,norm="g")
>
> ##And then the genefilter procedures,
> ## I want to flag the logratio(M) to NA according to the intensity of  
> both
> channels of a spot,and then select the genes if the
> ### sbsolute logratio(|M|) is grater than one.What can i do,what's the  
> code
> doing that?

The better way to do this in the limma framework is to set the weight  
for the spot to some small number (or 0).  A search of the bioconductor  
archives (here: http://files.protsuggest.org/cgi-bin/biocond.cgi) pulls  
up a couple of threads worth reading:

http://files.protsuggest.org/biocond/html/4342.html
http://files.protsuggest.org/biocond/html/3295.html

However, if you STILL really want to set them to NA after reading some  
more, you can do that simply like:

zraw.normrg$M[(zraw$R<200) | (zraw$G<200)] <- NA

Sean

Sean.