Hi BioC, I'm trying to do some differential expression analysis with one colour agilent data. What I have is 4 CY5 arrays in duplicate. They were both hybridised slightly differently and I want to check that the data generated was unaffected by the change. The working premise is that the differentially expressed genes between 2 arrays done under the first condition will be replicated under the second. These are the 8 slides. A9802 = SAMPLE1, A9811 = SAMPLE2 and so on. A9802 A9811 A9813 A9842 SAMPLE1 SAMPLE2 SAMPLE3 SAMPLE4 The data was loaded like this > RTestR <- read.maimages(targets2, columns = list(G = "rMeanSignal", Gb = "rBGUsed", R = "rMeanSignal", Rb = "rBGUsed"),annotation= c("Row", "Col", "FeatureNum", "ProbeUID","ControlType","ProbeName", "GeneName", "SystematicName"), wt.fun=myFlagFun) myFlagFun looks like this function(x) { #Weight only strongly positive spots 1, everything else 0 present <- x$rIsPosAndSignif == 1 probe <- x$ControlType == 0 manual <- x$IsManualFlag == 0 strong <- x$rIsWellAboveBG == 1 y <- as.numeric(present & probe & manual & strong) #Weight weak spots 0.5 weak <- strong == FALSE weak <- (present & probe & manual & weak) weak <- grep(TRUE,weak) y[weak] <- 0.5 #Weight flagged spots 0.5 sat <- x$rIsSaturated == 0 xdr <- x$rIsLowPMTScaledUp == 0 featureOL1 <- x$rIsFeatNonUnifOL == 0 featureOL2 <- x$rIsFeatPopnOL == 0 flagged <- (sat & xdr & featureOL1 & featureOL2) flagged <- grep(FALSE, flagged) good <- grep(TRUE, y==1) flagged <- intersect(flagged, good) y[flagged] <- 0.5 y } Combined the arrays > RTotal = cbind(RTestR[,1],RTestR[,2],RTestR[,3],RTestR[,4],RTestOriR[,1],RTestO ri R[,2],RTestOriR[,3],RTestOriR[,4]) I've checked to see I'm not just weighting everything out as 0 > summary(RTotal$weight>0) C:\\Program Files\\Agilent\\GeneSpring GX\\data\\CRX\\SingleChanneltest\\SAMPLE 1\\test01_251485017912_S01_GE2-v5_91_0806_1_1 Mode :logical FALSE:8796 TRUE :36219 C:\\Program Files\\Agilent\\GeneSpring GX\\data\\CRX\\SingleChanneltest\\SAMPLE 2\\test01_251485017912_S01_GE2-v5_91_0806_1_2 Mode :logical FALSE:10350 TRUE :34665 C:\\Program Files\\Agilent\\GeneSpring GX\\data\\CRX\\SingleChanneltest\\SAMPLE 3\\test01_251485017912_S01_GE2-v5_91_0806_1_3 Mode :logical FALSE:8785 TRUE :36230 C:\\Program Files\\Agilent\\GeneSpring GX\\data\\CRX\\SingleChanneltest\\SAMPLE 4\\test01_251485017912_S01_GE2-v5_91_0806_1_4 Mode :logical FALSE:7541 TRUE :37474 C:\\Program Files\\Agilent\\GeneSpring GX\\data\\CRX\\980\\test01_251485020980_S01_GE2-v5_91_0806_2_1_2 Mode :logical FALSE:10373 TRUE :34642 C:\\Program Files\\Agilent\\GeneSpring GX\\data\\CRX\\981\\981\\test01_251485020981_S02_GE2-v5_91_0806_2_1_1 Mode :logical FALSE:12630 TRUE :32385 C:\\Program Files\\Agilent\\GeneSpring GX\\data\\CRX\\981\\981\\test01_251485020981_S02_GE2-v5_91_0806_2_1_3 Mode :logical FALSE:12492 TRUE :32523 C:\\Program Files\\Agilent\\GeneSpring GX\\data\\CRX\\984\\984\\test01_251485020984_S02_GE2-v5_91_0806_2_1_2 Mode :logical FALSE:9496 TRUE :35519 Bg correct and normalise > RTotalbg = backgroundCorrect(RTotal, method="none") > RTRN <- normalizeBetweenArrays(RTotalbg$R, method="quantile") Comparing the slides to one another so get slide numbers from targets > f <- paste(RTotal$targets$SlideNumber,sep="") > f <- factor(f) > design <- model.matrix(~0+f) > colnames(design) <- levels(f) > design A9802 A9811 A9813 A9842 SAMPLE1 SAMPLE2 SAMPLE3 SAMPLE4 1 0 0 0 0 1 0 0 0 2 0 0 0 0 0 1 0 0 3 0 0 0 0 0 0 1 0 4 0 0 0 0 0 0 0 1 5 1 0 0 0 0 0 0 0 6 0 1 0 0 0 0 0 0 7 0 0 1 0 0 0 0 0 8 0 0 0 1 0 0 0 0 attr(,"assign") [1] 1 1 1 1 1 1 1 1 attr(,"contrasts") attr(,"contrasts")$f [1] "contr.treatment" >fit <- lmFit(RTRN, design) >cont.matrix <- makeContrasts(OneVOne="A9802-SAMPLE1",OneVTwo="A9802-A9811",TwoVOne="S AM PLE1-SAMPLE2",levels=design) The matrix seems to be doing what I want > cont.matrix Contrasts Levels OneVOne OneVTwo TwoVOne A9802 1 1 0 A9811 0 -1 0 A9813 0 0 0 A9842 0 0 0 SAMPLE1 -1 0 1 SAMPLE2 0 0 -1 SAMPLE3 0 0 0 SAMPLE4 0 0 0 > fit2 <- contrasts.fit(fit, cont.matrix) > fit2 <- eBayes(fit2) Error in ebayes(fit = fit, proportion = proportion, stdev.coef.lim = stdev.coef.lim) : No residual degrees of freedom in linear model fits I've found a post that says this error message occurs because all data is weighted out. I've checked the data after it is loaded, after backgroundCorrect and it does not appear to be. Beyond that I doesn't look like the normalizeBetweenArrays of RTotalbg$R RTRN has any weights. So I must not be setting up the design matrix correctly? Any and all clues as to where I'm going wrong greatly appreciated. Elliott Harrison This message has been scanned for viruses by BlackSpider...{{dropped:3}}

Hello Elliott: I am trying to understand the problem here. From the design: > A9802 A9811 A9813 A9842 SAMPLE1 SAMPLE2 SAMPLE3 SAMPLE4 > 1 0 0 0 0 1 0 0 0 > 2 0 0 0 0 0 1 0 0 > 3 0 0 0 0 0 0 1 0 > 4 0 0 0 0 0 0 0 1 > 5 1 0 0 0 0 0 0 0 > 6 0 1 0 0 0 0 0 0 > 7 0 0 1 0 0 0 0 0 > 8 0 0 0 1 0 0 0 0 I understand that there is one sample of A9802(#5) and another of SAMPLE1 (#1). In the contrasts, these two groups are compared (OneVOne): > makeContrasts(OneVOne="A9802-SAMPLE1",OneVTwo="A9802-A9811",TwoVOne= "SAM > PLE1-SAMPLE2",levels=design) > I guess that because of the number of samples in each group being one, it is not possible to calculate variance, and hence the error message. This is how I understood Gordon's earlier post (https://stat.ethz.ch/pipermail/bioconductor/2005-May/009056.html): ------------ The "no residual degrees of freedom" message occurs because you have filtered out so many spots (by setting the weight to 0) that you have no more than one spot left for any of the probes. Hence there is no replication left in your experiment. No estimate of variability can be made and no statistical analysis can be done. ------------- If anyone knows more clearly, please elaborate. Saroj > The matrix seems to be doing what I want > > >> cont.matrix > Contrasts > Levels OneVOne OneVTwo TwoVOne > A9802 1 1 0 > A9811 0 -1 0 > A9813 0 0 0 > A9842 0 0 0 > SAMPLE1 -1 0 1 > SAMPLE2 0 0 -1 > SAMPLE3 0 0 0 > SAMPLE4 0 0 0 > > >> fit2 <- contrasts.fit(fit, cont.matrix) fit2 <- eBayes(fit2) > Error in ebayes(fit = fit, proportion = proportion, stdev.coef.lim = > stdev.coef.lim) : No residual degrees of freedom in linear model fits > > > I've found a post that says this error message occurs because all data > is weighted out. I've checked the data after it is loaded, after > backgroundCorrect and it does not appear to be. Beyond that I doesn't look > like the normalizeBetweenArrays of RTotalbg$R RTRN has any weights. So I > must not be setting up the design matrix correctly? > > Any and all clues as to where I'm going wrong greatly appreciated. > > > > Elliott Harrison > > > > > > This message has been scanned for viruses by BlackSpider...{{dropped:3}} > > > _______________________________________________ > Bioconductor mailing list > Bioconductor at stat.math.ethz.ch > https://stat.ethz.ch/mailman/listinfo/bioconductor > Search the archives: > http://news.gmane.org/gmane.science.biology.informatics.conductor > >

Hi Saroj, I see so multiple arrays in each group are needed. So I'll need to do some simpler test between the 2 arrays? Any suggestions? Thanks Elliott -----Original Message----- From: smohapat@vbi.vt.edu [mailto:smohapat@vbi.vt.edu] Sent: Wednesday, November 07, 2007 12:07 PM To: elliott harrison Cc: bioconductor at stat.math.ethz.ch Subject: Re: [BioC] Problem running eBayes/lmFit Hello Elliott: I am trying to understand the problem here. From the design: > A9802 A9811 A9813 A9842 SAMPLE1 SAMPLE2 SAMPLE3 SAMPLE4 > 1 0 0 0 0 1 0 0 0 > 2 0 0 0 0 0 1 0 0 > 3 0 0 0 0 0 0 1 0 > 4 0 0 0 0 0 0 0 1 > 5 1 0 0 0 0 0 0 0 > 6 0 1 0 0 0 0 0 0 > 7 0 0 1 0 0 0 0 0 > 8 0 0 0 1 0 0 0 0 I understand that there is one sample of A9802(#5) and another of SAMPLE1 (#1). In the contrasts, these two groups are compared (OneVOne): > makeContrasts(OneVOne="A9802-SAMPLE1",OneVTwo="A9802-A9811",TwoVOne="S > AM > PLE1-SAMPLE2",levels=design) > I guess that because of the number of samples in each group being one, it is not possible to calculate variance, and hence the error message. This is how I understood Gordon's earlier post (https://stat.ethz.ch/pipermail/bioconductor/2005-May/009056.html): ------------ The "no residual degrees of freedom" message occurs because you have filtered out so many spots (by setting the weight to 0) that you have no more than one spot left for any of the probes. Hence there is no replication left in your experiment. No estimate of variability can be made and no statistical analysis can be done. ------------- If anyone knows more clearly, please elaborate. Saroj > The matrix seems to be doing what I want > > >> cont.matrix > Contrasts > Levels OneVOne OneVTwo TwoVOne > A9802 1 1 0 > A9811 0 -1 0 > A9813 0 0 0 > A9842 0 0 0 > SAMPLE1 -1 0 1 > SAMPLE2 0 0 -1 > SAMPLE3 0 0 0 > SAMPLE4 0 0 0 > > >> fit2 <- contrasts.fit(fit, cont.matrix) fit2 <- eBayes(fit2) > Error in ebayes(fit = fit, proportion = proportion, stdev.coef.lim = > stdev.coef.lim) : No residual degrees of freedom in linear model fits > > > I've found a post that says this error message occurs because all data > is weighted out. I've checked the data after it is loaded, after > backgroundCorrect and it does not appear to be. Beyond that I doesn't > look like the normalizeBetweenArrays of RTotalbg$R RTRN has any > weights. So I must not be setting up the design matrix correctly? > > Any and all clues as to where I'm going wrong greatly appreciated. > > > > Elliott Harrison > > > > > > This message has been scanned for viruses by > BlackSpider...{{dropped:3}} > > > _______________________________________________ > Bioconductor mailing list > Bioconductor at stat.math.ethz.ch > https://stat.ethz.ch/mailman/listinfo/bioconductor > Search the archives: > http://news.gmane.org/gmane.science.biology.informatics.conductor > > This message has been scanned for viruses by BlackSpider...{{dropped:3}}