Hi everyone, I have an interesting phenomenon in some microarray data, and wondered if anyone else has seen anything like it. It's 2-color data, comparing mutant vs. wildtype, 2 replicates plus dye-swaps for a total of 4 arrays. The 'dye-swaps', instead of being negatively correlated in M-values are instead strongly positively correlated, even after within-array normalization. I triple checked to make sure I didn't have the phenotypic info wrong, but all of the arrays are positively correlated, which leads me to believe that dye-swapping wasn't actually done. If you analyze as if it were a dye-swap experiment, several thousands of genes still show a dye-effect, whereas only dozens of genes show a MUvWT effect. My question: is it possible that any dye-effects could be so strong, even after within-array normalization, and treatment differences so small that the arrays could be dye-swaps but still show a positive correlation in M-values? Or is it more likely that dye-swapping wasn't actually done? I've tried to look at other dye-swapped data, but everything I have has large treatment differences. The PI already has the manuscript written, and just came to me to 'confirm' their analysis, so I want to be pretty positive before I tell them their work may have been wasted (of course, they may still decide to ignore me...) Thanks, Jenny Jenny Drnevich, Ph.D. Functional Genomics Bioinformatics Specialist W.M. Keck Center for Comparative and Functional Genomics Roy J. Carver Biotechnology Center University of Illinois, Urbana-Champaign 330 ERML 1201 W. Gregory Dr. Urbana, IL 61801 USA ph: 217-244-7355 fax: 217-265-5066 e-mail: drnevich at uiuc.edu

Hi Jenny, There are a couple of other things you can check to make sure you have the correct orientation: - if this is a mutant vs. control, the investigator probably know about some upregulated/downregulated genes which caracterize the mutant. You can check the sign of the M values for probes corresponding to these genes to determine the correct orientation of the arrays. - If you have access to the actual image of the arrays, differentially expressed probes should show up with different colors on dye-swapped arrays. - Also, if the investigator already checked the expression of some of the genes using another method (like taqman), you could use this as a "true" value and check which arrays have the correct dye orientation. Regards, Agnes ________________________________ From: bioconductor-bounces@stat.math.ethz.ch on behalf of Jenny Drnevich Sent: Mon 4/30/2007 11:36 AM To: bioconductor at stat.math.ethz.ch Subject: [BioC] dye effects stronger than dye-swaps? Hi everyone, I have an interesting phenomenon in some microarray data, and wondered if anyone else has seen anything like it. It's 2-color data, comparing mutant vs. wildtype, 2 replicates plus dye-swaps for a total of 4 arrays. The 'dye-swaps', instead of being negatively correlated in M-values are instead strongly positively correlated, even after within-array normalization. I triple checked to make sure I didn't have the phenotypic info wrong, but all of the arrays are positively correlated, which leads me to believe that dye-swapping wasn't actually done. If you analyze as if it were a dye-swap experiment, several thousands of genes still show a dye-effect, whereas only dozens of genes show a MUvWT effect. My question: is it possible that any dye-effects could be so strong, even after within-array normalization, and treatment differences so small that the arrays could be dye-swaps but still show a positive correlation in M-values? Or is it more likely that dye-swapping wasn't actually done? I've tried to look at other dye-swapped data, but everything I have has large treatment differences. The PI already has the manuscript written, and just came to me to 'confirm' their analysis, so I want to be pretty positive before I tell them their work may have been wasted (of course, they may still decide to ignore me...) Thanks, Jenny Jenny Drnevich, Ph.D. Functional Genomics Bioinformatics Specialist W.M. Keck Center for Comparative and Functional Genomics Roy J. Carver Biotechnology Center University of Illinois, Urbana-Champaign 330 ERML 1201 W. Gregory Dr. Urbana, IL 61801 USA ph: 217-244-7355 fax: 217-265-5066 e-mail: drnevich at uiuc.edu _______________________________________________ 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

On Monday 30 April 2007 14:36, Jenny Drnevich wrote: > Hi everyone, > > I have an interesting phenomenon in some microarray data, and > wondered if anyone else has seen anything like it. It's 2-color data, > comparing mutant vs. wildtype, 2 replicates plus dye-swaps for a > total of 4 arrays. The 'dye-swaps', instead of being negatively > correlated in M-values are instead strongly positively correlated, > even after within-array normalization. I triple checked to make sure > I didn't have the phenotypic info wrong, but all of the arrays are > positively correlated, which leads me to believe that dye-swapping > wasn't actually done. If you analyze as if it were a dye-swap > experiment, several thousands of genes still show a dye-effect, > whereas only dozens of genes show a MUvWT effect. > > My question: is it possible that any dye-effects could be so strong, > even after within-array normalization, and treatment differences so > small that the arrays could be dye-swaps but still show a positive > correlation in M-values? Or is it more likely that dye-swapping > wasn't actually done? I've tried to look at other dye-swapped data, > but everything I have has large treatment differences. The PI already > has the manuscript written, and just came to me to 'confirm' their > analysis, so I want to be pretty positive before I tell them their > work may have been wasted (of course, they may still decide to ignore > me...) Jenny, Just a quick question--were the samples amplified? Sean

And then again, there is the dye degradation problem. If one dye severely degrades, then you will get a strong positive correlation. Try plotting R vs R and G vs G, instead of M vs M. --Naomi At 03:56 PM 4/30/2007, Paquet, Agnes wrote: >Hi Jenny, > >There are a couple of other things you can check to make sure you >have the correct orientation: >- if this is a mutant vs. control, the investigator probably know >about some upregulated/downregulated genes which caracterize the >mutant. You can check the sign of the M values for probes >corresponding to these genes to determine the correct orientation of >the arrays. >- If you have access to the actual image of the arrays, >differentially expressed probes should show up with different colors >on dye-swapped arrays. >- Also, if the investigator already checked the expression of some >of the genes using another method (like taqman), you could use this >as a "true" value and check which arrays have the correct dye orientation. > >Regards, > >Agnes > >________________________________ > >From: bioconductor-bounces at stat.math.ethz.ch on behalf of Jenny Drnevich >Sent: Mon 4/30/2007 11:36 AM >To: bioconductor at stat.math.ethz.ch >Subject: [BioC] dye effects stronger than dye-swaps? > > > >Hi everyone, > >I have an interesting phenomenon in some microarray data, and >wondered if anyone else has seen anything like it. It's 2-color data, >comparing mutant vs. wildtype, 2 replicates plus dye-swaps for a >total of 4 arrays. The 'dye-swaps', instead of being negatively >correlated in M-values are instead strongly positively correlated, >even after within-array normalization. I triple checked to make sure >I didn't have the phenotypic info wrong, but all of the arrays are >positively correlated, which leads me to believe that dye-swapping >wasn't actually done. If you analyze as if it were a dye-swap >experiment, several thousands of genes still show a dye-effect, >whereas only dozens of genes show a MUvWT effect. > >My question: is it possible that any dye-effects could be so strong, >even after within-array normalization, and treatment differences so >small that the arrays could be dye-swaps but still show a positive >correlation in M-values? Or is it more likely that dye-swapping >wasn't actually done? I've tried to look at other dye-swapped data, >but everything I have has large treatment differences. The PI already >has the manuscript written, and just came to me to 'confirm' their >analysis, so I want to be pretty positive before I tell them their >work may have been wasted (of course, they may still decide to ignore me...) > >Thanks, >Jenny > >Jenny Drnevich, Ph.D. > >Functional Genomics Bioinformatics Specialist >W.M. Keck Center for Comparative and Functional Genomics >Roy J. Carver Biotechnology Center >University of Illinois, Urbana-Champaign > >330 ERML >1201 W. Gregory Dr. >Urbana, IL 61801 >USA > >ph: 217-244-7355 >fax: 217-265-5066 >e-mail: drnevich at uiuc.edu > >_______________________________________________ >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 > >_______________________________________________ >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 Naomi S. Altman 814-865-3791 (voice) Associate Professor Dept. of Statistics 814-863-7114 (fax) Penn State University 814-865-1348 (Statistics) University Park, PA 16802-2111