Question: Two channel 4x44k array analysis in Limma
0
gravatar for nemo566
3.2 years ago by
nemo5660
nemo5660 wrote:

Hi there, I am trying to analyze two channel Agilent 4x44k microarray by Limma in R. So far I used Limma guide to analyze my data. Unfortunately, I don't have any DE after analysis. Here is my code:

targets<-readTargets("targets.txt")
RG <- read.maimages(targets, path="directory",source="agilent.median")'
RG <- backgroundCorrect(RG, method="normexp", offset=10)
MA <- normalizeWithinArrays(RG, method="loess")

Adjusted p value at the beginning was much more higher than 0.05. So I decided to use this command for cutting of my spike in probes:

MA2<-MA[MA$genes$ControlType==0,]

Then, I came back to standard code:

MA.avg <- avereps(MA2, ID=MA2$genes$ProbeName)
design <- modelMatrix(targets, ref="cycle")
fit <- lmFit(MA.avg, design)
fit2 <- eBayes(fit)
output <- topTable(fit2, adjust="fdr", coef="Pregnancy", number=100000)
write.table(output, file="Results.txt", sep="\t", quote=FALSE)

Still I don't have any DE genes. The least adjusted p value=0.3264. I found out that there is one more option. I can cut off low expressed genes from my list of 45k genes just after normalization step. Here is the code which let me to do that considering sd:

sds<-apply(MA, 1,sd)
cutsd<-quantile(sds, 0.95)
ind<-sds>cutsd
MA.sd<-MA[ind,]

Is it a good one? What does the quantile, 1 and 0.95 value mean? I shouldn't change it, right? If anyone could help me...

I know that there is another option- to do more arrays. But firstly I want to check all possibilities with these which I've done so far. Thanks, a lot! ;)

ADD COMMENTlink modified 3.2 years ago • written 3.2 years ago by nemo5660
Please log in to add an answer.

Help
Access

Use of this site constitutes acceptance of our User Agreement and Privacy Policy.
Powered by Biostar version 2.3.0
Traffic: 1727 users visited in the last hour