Thanks Marcus ! Appreciate it -Rich On Mar 17, 2012, at 5:31 PM, Marcus Davy <mdavy86@gmail.com> wrote: > When you run avereps() on an MAList you are calling > > avereps.MAList <- > function (x, ID = NULL) > { > if (is.null(ID)) { > ID <- x$genes$ID > if (is.null(ID)) > ID <- rownames(x) > if (is.null(ID)) > stop("Cannot find probe IDs") > } > y <- x > y$M <- avereps(x$M, ID = ID) > y$A <- avereps(x$A, ID = ID) > other <- names(x$other) > for (a in other) y$other[[a]] <- avereps(x$other[[a]], ID = ID) > y$weights <- avereps(x$weights, ID = ID) > y$genes <- x$genes[!duplicated(ID), ] > y$printer <- NULL > y > } > <environment: namespace:limma=""> > > If you transform from a MAList to a RGList with RG.MA > > RG.MA > function (object) > { > object$R <- 2^(object$A + object$M/2) > object$G <- 2^(object$A - object$M/2) > object$M <- NULL > object$A <- NULL > new("RGList", unclass(object)) > } > <environment: namespace:limma=""> > > The RG intensities in your case are normalized and back transformed. > > The dyes are; > Cy3 => Green dye > Cy5 => Red dye. > > Your expression on the first 12 probes is effectively calculating M > > log2(E2.avg_RG$R[,1:12]/E2.avg_RG$G[,1:12]) > > or > > log2(E2.avg_RG$R[,1:12]) - log2(E2.avg_RG$G[,1:12]) > > where M = log2(R) - log2(G) > > You could verify this with a sanity check, something like; > > Cy5Cy3_RvsG <- log2(E2.avg_RG$R[,1:12]/E2.avg_RG$G[,1:12]) > identical(E2.avg$M[1:12,] , Cy5Cy3_RvsG) > > > Marcus > > > On Sun, Mar 18, 2012 at 11:19 AM, Richard Green <greener@uw.edu> wrote: > If I load and normalize some two channel arrays > > library(limma) > > setwd("/vol04/microarray/vineet_study/run_data2") > > files <- dir(pattern="*.txt") > > RG <- read.maimages(files=dir(),source="agilent", > columns=list(G="gMeanSignal",Gb="gBGMedianSignal",R="rMeanSignal",Rb ="rBGMedianSignal"), > annotation = c("Row", "Col","FeatureNum", "GeneName", > "ControlType","ProbeName")) > > targets <- readTargets("/vol04/microarray/vineet_study/targets.txt") > > RG<- backgroundCorrect(RG, method="none") > > MA <- normalizeWithinArrays(RG, method="loess") > > E2.avg <- avereps(MA, ID=MA$genes$ProbeName) > > and I want to generate ratios of Cy5/Cy3 and Cy3/Cy5 > > I could convert my MA back to and RG > > E2.avg_RG <- RG.MA(E2.avg) > > if I do that it will create two unlogged channels again. > > question 1 are the intensities still normalized? which channel is Cy3 and > Cy5? > > Would the following generate a log ratio of my normalized data(for the > first 12 samples)? > > Cy5Cy3_RvsG <- log2(E2.avg_RG$R[,1:12]/E2.avg_RG$G[,1:12]) > > Any advice folks have is appreciated. > > Thanks > > -Rich > > [[alternative HTML version deleted]] > > _______________________________________________ > Bioconductor mailing list > Bioconductor@r-project.org > https://stat.ethz.ch/mailman/listinfo/bioconductor > Search the archives: http://news.gmane.org/gmane.science.biology.informatics.conductor > [[alternative HTML version deleted]]

Checking slide samples would be Cy5Cy3_RvsG <- log2(E2.avg_RG$R[,1:12]/E2.avg_RG$G[,1:12]) identical(E2.avg$M[,1:12] , Cy5Cy3_RvsG) Marcus On Sun, Mar 18, 2012 at 1:31 PM, Marcus Davy <mdavy86@gmail.com> wrote: > When you run avereps() on an MAList you are calling > > avereps.MAList <- > function (x, ID = NULL) > { > if (is.null(ID)) { > ID <- x$genes$ID > if (is.null(ID)) > ID <- rownames(x) > if (is.null(ID)) > stop("Cannot find probe IDs") > } > y <- x > y$M <- avereps(x$M, ID = ID) > y$A <- avereps(x$A, ID = ID) > other <- names(x$other) > for (a in other) y$other[[a]] <- avereps(x$other[[a]], ID = ID) > y$weights <- avereps(x$weights, ID = ID) > y$genes <- x$genes[!duplicated(ID), ] > y$printer <- NULL > y > } > <environment: namespace:limma=""> > > If you transform from a MAList to a RGList with RG.MA > > RG.MA > function (object) > { > object$R <- 2^(object$A + object$M/2) > object$G <- 2^(object$A - object$M/2) > object$M <- NULL > object$A <- NULL > new("RGList", unclass(object)) > } > <environment: namespace:limma=""> > > The RG intensities in your case are normalized and back transformed. > > The dyes are; > Cy3 => Green dye > Cy5 => Red dye. > > Your expression on the first 12 probes is effectively calculating M > > log2(E2.avg_RG$R[,1:12]/E2.avg_RG$G[,1:12]) > > or > > log2(E2.avg_RG$R[,1:12]) - log2(E2.avg_RG$G[,1:12]) > > where M = log2(R) - log2(G) > > You could verify this with a sanity check, something like; > > Cy5Cy3_RvsG <- log2(E2.avg_RG$R[,1:12]/E2.avg_RG$G[,1:12]) > identical(E2.avg$M[1:12,] , Cy5Cy3_RvsG) > > > Marcus > > > On Sun, Mar 18, 2012 at 11:19 AM, Richard Green <greener@uw.edu> wrote: > >> If I load and normalize some two channel arrays >> >> library(limma) >> >> setwd("/vol04/microarray/vineet_study/run_data2") >> >> files <- dir(pattern="*.txt") >> >> RG <- read.maimages(files=dir(),source="agilent", >> >> columns=list(G="gMeanSignal",Gb="gBGMedianSignal",R="rMeanSignal",R b="rBGMedianSignal"), >> annotation = c("Row", "Col","FeatureNum", "GeneName", >> "ControlType","ProbeName")) >> >> targets <- readTargets("/vol04/microarray/vineet_study/targets.txt") >> >> RG<- backgroundCorrect(RG, method="none") >> >> MA <- normalizeWithinArrays(RG, method="loess") >> >> E2.avg <- avereps(MA, ID=MA$genes$ProbeName) >> >> and I want to generate ratios of Cy5/Cy3 and Cy3/Cy5 >> >> I could convert my MA back to and RG >> >> E2.avg_RG <- RG.MA(E2.avg) >> >> if I do that it will create two unlogged channels again. >> >> question 1 are the intensities still normalized? which channel is Cy3 and >> Cy5? >> >> Would the following generate a log ratio of my normalized data(for the >> first 12 samples)? >> >> Cy5Cy3_RvsG <- log2(E2.avg_RG$R[,1:12]/E2.avg_RG$G[,1:12]) >> >> Any advice folks have is appreciated. >> >> Thanks >> >> -Rich >> >> [[alternative HTML version deleted]] >> >> _______________________________________________ >> Bioconductor mailing list >> Bioconductor@r-project.org >> https://stat.ethz.ch/mailman/listinfo/bioconductor >> Search the archives: >> http://news.gmane.org/gmane.science.biology.informatics.conductor >> > > [[alternative HTML version deleted]]