Search
Question: Limma: how to obtain P values using method "global" for multiple testing
0
gravatar for Mark Cruickshank
6.3 years ago by
Mark Cruickshank10 wrote:
Dear Eduardo, I understand that "method" relates to how contrasts are compared. There is a description in these posts: https://stat.ethz.ch/pipermail/bioconductor/2008-January/020646.html https://mailman.stat.ethz.ch/pipermail/bioconductor/2012-May/045488.ht ml You will see that "separate" method in decideTests will give the same result as topTable. This is a very easy way to interpret the results. P-vals from your two contrasts from topTable will correspond to the correlated (1) or anti-correlated (-1) probes. You can get a sense if your contrasts are correlated by displaying your decideTests results Venn or with heatDiagram in limma. i'm very new to arrays & interested if others have different views. This is something you could try. cont.matrix <- makeContrasts( 'contA'=(TreatmentA - Control), 'contB'=(TreatmentB - Control), levels=design) fit<-lmFit(M,design) fit2 <- contrasts.fit(fit, cont.matrix) fit2 <- eBayes(fit2) DECIDE<-decideTests(fit2,method="separate",adjust.method="BH",p.value= 0.05,lfc=0) contA<-topTable(fit2,coef=1,adjust.method="BH",number=480000,sort.by=" B") contB<-topTable(fit2,coef=2,adjust.method="BH",number=480000,sort.by=" B") DECIDE<-decideTests(fit2,method="separate",adjust.method="BH",p.value= 0.05,lfc=0) heatDiagram(DECIDE,coef=fit2$coef) vennDiagram(DECIDE) Cheers, Mark -----Original Message----- From: bioconductor-bounces@r-project.org on behalf of Ribeiro,Eduardo de Souza Sent: Sat 7/21/2012 12:02 PM To: bioconductor@r-project.org Subject: [BioC] Limma: how to obtain P values using method "global" for multiple testing Hi everyone, I am analyzing a Affymetrix microarray data set using the Limma package. According to my experimental design, I understand that my contrasts are closely related and that I should use the "global" method for multiple testing. My question is: Is there a way to obtain P-values using the "global" method for multiple testing? The decideTest function allows to specify the method "global" but just give the matrix of the results and not the P-values. And if I understood correctly, the topTable function does not allow to specify the method for multiple testing and uses the "separate" method as default. Comparing the decideTest matrices, I realize that the results of "separate" and "global" methods are slightly different and I am concerned that this difference may change the final interpretation of my study. Thank you very much. Best regards Eduardo Ribeiro PhD student University of Florida [[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 ______________________________________________________________________ This email has been scanned by the Symantec Email Security.cloud service. For more information please visit http://www.symanteccloud.com If you have any question, please contact MCRI IT Helpdesk for further assistance. ______________________________________________________________________ ______________________________________________________________________ This email has been scanned by the Symantec Email Security.cloud service. For more information please visit http://www.symanteccloud.com ______________________________________________________________________ [[alternative HTML version deleted]]
ADD COMMENTlink modified 6.3 years ago by Ribeiro,Eduardo de Souza30 • written 6.3 years ago by Mark Cruickshank10
0
gravatar for Ribeiro,Eduardo de Souza
6.3 years ago by
Hi Mark, Thank you very much for your reply. I understand that I can use the "separate" method to obtain the same results in the decideTest and topTable. I was just wondering if it was possible to obtain P-values using the "global" method. I am not sure if that makes sense. Maybe it is just my misunderstanding of the statistical concepts involving multiple testing for contrasts. Cheers Eduardo Ribeiro PhD student University of Florida ________________________________ From: Mark Cruickshank [mark.cruickshank@mcri.edu.au] Sent: Saturday, July 21, 2012 3:16 AM To: Ribeiro,Eduardo de Souza; bioconductor@r-project.org Subject: RE: [BioC] Limma: how to obtain P values using method "global" for multiple testing Dear Eduardo, I understand that "method" relates to how contrasts are compared. There is a description in these posts: https://stat.ethz.ch/pipermail/bioconductor/2008-January/020646.html https://mailman.stat.ethz.ch/pipermail/bioconductor/2012-May/045488.ht ml You will see that "separate" method in decideTests will give the same result as topTable. This is a very easy way to interpret the results. P-vals from your two contrasts from topTable will correspond to the correlated (1) or anti-correlated (-1) probes. You can get a sense if your contrasts are correlated by displaying your decideTests results Venn or with heatDiagram in limma. i'm very new to arrays & interested if others have different views. This is something you could try. cont.matrix <- makeContrasts( 'contA'=(TreatmentA - Control), 'contB'=(TreatmentB - Control), levels=design) fit<-lmFit(M,design) fit2 <- contrasts.fit(fit, cont.matrix) fit2 <- eBayes(fit2) DECIDE<-decideTests(fit2,method="separate",adjust.method="BH",p.value= 0.05,lfc=0) contA<-topTable(fit2,coef=1,adjust.method="BH",number=480000,sort.by=" B") contB<-topTable(fit2,coef=2,adjust.method="BH",number=480000,sort.by=" B") DECIDE<-decideTests(fit2,method="separate",adjust.method="BH",p.value= 0.05,lfc=0) heatDiagram(DECIDE,coef=fit2$coef) vennDiagram(DECIDE) Cheers, Mark -----Original Message----- From: bioconductor-bounces@r-project.org on behalf of Ribeiro,Eduardo de Souza Sent: Sat 7/21/2012 12:02 PM To: bioconductor@r-project.org Subject: [BioC] Limma: how to obtain P values using method "global" for multiple testing Hi everyone, I am analyzing a Affymetrix microarray data set using the Limma package. According to my experimental design, I understand that my contrasts are closely related and that I should use the "global" method for multiple testing. My question is: Is there a way to obtain P-values using the "global" method for multiple testing? The decideTest function allows to specify the method "global" but just give the matrix of the results and not the P-values. And if I understood correctly, the topTable function does not allow to specify the method for multiple testing and uses the "separate" method as default. Comparing the decideTest matrices, I realize that the results of "separate" and "global" methods are slightly different and I am concerned that this difference may change the final interpretation of my study. Thank you very much. Best regards Eduardo Ribeiro PhD student University of Florida [[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 ______________________________________________________________________ This email has been scanned by the Symantec Email Security.cloud service. For more information please visit http://www.symanteccloud.com If you have any question, please contact MCRI IT Helpdesk for further assistance. ______________________________________________________________________ ______________________________________________________________________ This email has been scanned by the Symantec Email Security.cloud service. For more information please visit http://www.symanteccloud.com ______________________________________________________________________ [[alternative HTML version deleted]]
ADD COMMENTlink written 6.3 years ago by Ribeiro,Eduardo de Souza30
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 2.2.0
Traffic: 405 users visited in the last hour