[BioC] Remove some genes from analysis in limma package? Is it possible?

Marcelo Luiz de Laia mlaia at fcav.unesp.br
Sat Oct 25 14:39:35 MEST 2003

James Wettenhall, kindly, sent me some explanations on how to accomplish the
analyses without some genes.

However, I tried to accomplish the procedure and I didn't get.

I am sure that I didn't get because I don't have experience with the
softwere R, because I believe that my doubt is related with the softwere R.
Really, I don't know anything about R, I am learning how to work with R,
with the packages of Bioconductor and with microarrays analyses now (and
english too)! Everything this concomitant!

I also know that it is very annoying  (importunate) to be answering to basic
questions, of beginners. But, to me it doesn't remain another option and I
apologize with all you for proceeding of this sort things out.

James wrote me that it was possible to accomplish a subset of the sample
using one of the following resources:

MAnocontrols < - MA[RG$genes [, "ID"] != "control",]


status < - rep ("gene",nrow(RG$genes))
status[grep ("control",RG$genes [, "ID"])] < - "control"
MAnocontrols2 < - MA[status!= "control",]

I tried to execute the analyses and I didn't get. Errors happened.

At the end of this message I am enclosing the steps that I proceeded to
accomplish the analysis.

In my file of data for the analysis (Imagene ou put file) have some genes
which I would not like that they were considered in the analysis. I would
like to execute the normalizations without these genes, because the same
ones are interfering negatively in them analysis.

How could I accomplish these analyses?

thanks a lot


> slides<-cbind(RegianeSample$FileNameCy3, RegianeSample$FileNameCy5)
> RG<-read.maimages(slides, source="imagene")
Read header information
Read lamina1_1_selected.txt
Read lamina1_0_selected.txt
Read lamina2_1_selected.txt
Read lamina2_0_selected.txt
Read lamina3_1_selected.txt
Read lamina3_0_selected.txt
> layout<-RG$printer
> genes<-RG$genes
> MAsemplaca <- MA[RG$genes[,"Gene ID"]!="Sem Placa",]
Error in "[.MAList"(MA, RG$genes[, "Gene ID"] != "Sem Placa", ) :
        (subscript) logical subscript too long
> status <- rep("gene",nrow(RG$genes))
> status[grep("Sem Placa",RG$genes[,"Gene ID"])] <- "Sem Placa"
Error in grep(pattern, x, ignore.case, extended, value, fixed) :
        invalid argument
> MAraw <- MA.RG(RG)
> MA <- normalizeWithinArrays(RG,layout)
> MAsemplaca <- MA[RG$genes[,"Gene ID"]!="Sem Placa",]
> design=c(1,-1,-1)
> fit <- lm.series(MA$M,design)
> eb <- ebayes(fit)
> qqt(eb$t,df=fit$df+eb$df,pch=16,cex=0.1)
> options(digits=3)
> toptable(number=50, genelist=genes, fit=fit, eb=eb, adjust="fdr")
        Field Meta.Row Meta.Column Row Column       Gene.ID
3769 XDM/BCYE        3           1  30      1     Sem Placa

"Sem Placa" is the first gene in the top 50!

-----Original Message-----
From: James Wettenhall [mailto:wettenhall at wehi.edu.au]
Sent: sexta-feira, 24 de outubro de 2003 02:46
To: Marcelo Luiz de Laia
Cc: bioconductor at stat.math.ethz.ch
Subject: Re: [BioC] Remove some genes from analysis in limma package? Is
it possible?


RG and MA objects in limma are subsettable, so you can select
the genes you want, e.g.

MAnocontrols <- MA[RG$genes[,"ID"]!="control",]

or you might want to define a "status" vector, as used in the
plotMA function:

status <- rep("gene",nrow(RG$genes))
status[grep("control",RG$genes[,"ID"])] <- "control"
MAnocontrols2 <- MA[status!="control",]

The "grep" method allows you to select multiple spots and give
them the same spot type status, e.g. Calibration_Control_1 and
Calibration_Control_2 could all be given the spot type
"calibration" and then excluded from the analysis.

You could replace "control" above by "ZZZZXXX21".

Hope this helps,

James Wettenhall                                  Tel: (+61 3) 9345 2629
Division of Genetics and Bioinformatics           Fax: (+61 3) 9347 0852
The Walter & Eliza Hall Institute         E-mail: wettenhall at wehi.edu.au
 of Medical Research,                     Mobile: (+61 / 0 ) 438 527 921
1G Royal Parade,
Parkville, Vic 3050, Australia

Incoming mail is certified Virus Free.


More information about the Bioconductor mailing list