Hello all,

I have some custom Agilent arrays with 3600 spots representing  
functional genes.

In each of the RNA samples the same amount of exogenous RNA was spiked  
in as a control, these are the "Arab" spots in the script below.

I want to compare some treatments that were never hybridized on the  
same slide, therefore I want to do seperate channel analysis of two  
channel arrays.

The script below - follows from the user's guide.  However,  I am  
unsure about the most appropriate way to nomalize my data between the  
arrays.

When I looked at the MA plots the "Arab" spots are often  
differentially expressed.

Is there a way that I can normalize between arrays using a modified  
quantile, where all of the points would be normalized based on  trying  
to obtain equal distributions of a subset of spots on the array (ie.  
the "Arab" control spots)?

Thank you,

alison


 >>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>
targets<-readTargets("EPBDIExprTargets.txt")
RG<-read.maimages(targets$FileName,source='genepix')
spottypes<-readSpotTypes("SpotTypesEP.txt")
RGnm<-backgroundCorrect(RG,method='normexp')
RGnm$genes$Status<-controlStatus(spottypes,RG)
ControlSpots<-grep("Arab",RGnm$genes$Status)
w<-modifyWeights(array(1,dim(RGnm)), RGnm$genes$Status,  
c("All","Arab"), c(1,2))
MAnmw<-normalizeWithinArrays(RGnm,weights=w)
MAnmwq<-normalizeBetweenArrays(MAnmw, method="quantile")
targets2<-targetsA2C(targets)
u<-unique(targets2$Target)
f<-factor(targets2$Target, levels=u)
design<-model.matrix(~0+f)
colnames(design)<-u
corfit<-intraspotCorrelation(MAnmwq, design)
fit<-lmscFit(MAnmwq, design, correlation=corfit$consensus)
cont.matrix<-makeContrasts("PCE-TCE","PCE-cisDCE","PCE-VC", "PCE- 
lactate","TCE-cisDCE","TCE-VC", "TCE-lactate","cisDCE-VC", "cisDCE- 
lactate","VC-lactate",levels=design)
fit2<-contrasts.fit(fit, cont.matrix)
fit2<-eBayes(fit2)
plotMA(fit2)

---------------------------------------------------------
Alison Waller  Ph.D
alison.waller@utoronto.ca





	[[alternative HTML version deleted]]