Search
Question: Limma and drug treatment
0
14.2 years ago by
Nolwenn Le Meur30 wrote:
Dear all, I would like to found out if there is an interaction between treatments (Amines and Betabloc) on gene expression. I would like to known if some genes are differentially expressed by one of the drugs(Betabloc,Amines) or their combination (Betabloc:Amines). Reading the Bioconductor FAQ I think that these questions can be answered by Limma but I am not sure of what I get for result with the following script. The design matrix only gives me the first topres (diffrentially expressed) genes for the clinical parameter "Betabloc". How can I get the data for the Betabloc:Amines relationship. Do I have to write a specific design matrix like design<-model.matrix(~Betabloc:Amines-1,data=pData(clinicFull)) ? Does the following contrast matrix give me the genes differentially expressed between patients taking "Betabloc" and the one taking "Amines" ? Thank you in advance,Best regards, Nolwenn Le Meur ##Code VD<-read.table("VD.txt",header=TRUE,sep="\t",row.names=1) clinicFull<-read.phenoData("clinic.txt",header=TRUE,sep="\t",row.name s=1) eset<-new("exprSet",exprs=VD1,phenoData= clinicFull) eset #Expression Set (exprSet) with # 571 genes # 94 samples # phenoData object with 18 variables and 94 cases # varLabels # SampleID: read from file # CODEBDD: read from file # PAT: read from file # SEXE: read from file # Tabac: read from file # Diabete: read from file # HTA: read from file # Ifam: read from file # Dyslipid: read from file # IEC: read from file # Betabloc: read from file # Amines: read from file # Inh: read from file # Antialdo: read from file # duir: read from file # statine: read from file # amio: read from file # digi: read from file ##Model Betabloc+Amines+Betabloc:Amines design<-model.matrix(~Betabloc*Amines-1,data=pData(clinicFull)) fit <- lmFit(eset, design) eb<-ebayes(fit) topres<-toptable(number=571,genelist=as.character(row.names(VD)),fit=f it,eb= eb,adjust="holm") length(which(topres$P.Value<0.01)) cont.matrix<-makeContrasts(Betabloc-Amines,levels=design) fit1<-contrasts.fit(fit,cont.matrix) eb1<-ebayes(fit1) topres1<-toptable(number=571,genelist=as.character(row.names(VD)),fit= fit1,e b=eb1,adjust="holm") length(which(topres1$P.Value<0.01)) ******************************************** Nolwenn Le Meur INSERM U533 Facult? de m?decine 1, rue Gaston Veil 44035 Nantes Cedex 1 France Tel: (+33)-2-40-41-29-86 (office) (+33)-2-40-41-28-44 (secretary) Fax: (+33)-2-40-41-29-50 mail: nolwenn.lemeur@nantes.inserm.fr
modified 14.2 years ago by huj@sciosinc.com20 • written 14.2 years ago by Nolwenn Le Meur30
0
14.2 years ago by
Sean Davis21k
United States
Sean Davis21k wrote:
> > #Expression Set (exprSet) with > # 571 genes > # 94 samples > # phenoData object with 18 variables and 94 cases > # varLabels > > ##Model Betabloc+Amines+Betabloc:Amines > design<-model.matrix(~Betabloc*Amines-1,data=pData(clinicFull)) > fit <- lmFit(eset, design) > eb<-ebayes(fit) > topres<- > toptable(number=571,genelist=as.character(row.names(VD)),fit=fit,eb= > eb,adjust="holm") > length(which(topres$P.Value<0.01) My guess is that your eb contains information for each of the coefficients. You need to specify the coef parameter (1, 2, or 3) to get information about the other coefficients (the default coefficient is 1). Along the same lines, if you do: dim(eb$p.value) my guess is that you have 3 columns. Check to see if that is the case. See the documentation for topTable for using the coef parameter. > cont.matrix<-makeContrasts(Betabloc-Amines,levels=design) > fit1<-contrasts.fit(fit,cont.matrix) > eb1<-ebayes(fit1) > topres1<- > toptable(number=571,genelist=as.character(row.names(VD)),fit=fit1,e > b=eb1,adjust="holm") > length(which(topres1\$P.Value<0.01)) > This looks right to me given what is above. Sean
0
14.2 years ago by
huj@sciosinc.com20 wrote:
Hi all, I would like to try LPE for cDNA microarray data analysis. Can anyone please tell me how to install this package? The package list for getBioC() does not seem to include this package. Also, since I am fairly new to this field, is this a good test for identifying significant differentially expressed genes in cDNA microarray experiments? what are the alternatives? Thank you very much in advance, Jeff [[alternative HTML version deleted]]