Hi, In Chapter 8.1.2 in Limma users guide there is a description how to detecting the Dye effect. The example there describes an experiment of Wt vs Mut with two dye swaps replicates as follow: FileName Cy3 Cy5 File1 wt mu File2 mu wt File3 wt mu File4 mu wt Let's say that I found a list of gene which are significant due to mutant effect and due to dye effect as well. Can I only ignore them or can I correct the dye effect? I guess that it depends how the dye swaps replicates was prepared. If the dye replicates are technical replicates using the block design is enough to correct dye effect. If dye replicates are also biological replicates they night also represent a batch effect. That is all first replicates was sent in one day and the second replicates sent in another day, which this difference by itself without dye swap may be a source for variation. The second option of dye swap preparation may be corrected by ANCOVA (?). Is possible to do it with LIMMA? I know that a question about using ANCOVA with LIMMA arose by Naomi Altman but this discussion was beyond my knowledge. So back to my questions: Is correcting the dye effect is possible in LIMMA? Is ANCOVA is a solution? and HOW? I hope these questions are in the focus of that mailing list, Thanks Ron

The simplest way to detect dye effects is to plot M vs M on dye-swapped arrays. These should have negative correlation. If there is an X on the plot, you have lots of dye bias. The main causes I have seen for this are: 1) bad dye batches 2) ozone degradation of Cy5 Personally, I would not trust a statistical correction under these circumstances. On the other hand, if only a few genes are affected (as you would detect from a statistical rather than graphical analysis), probably correcting for the dye effect is OK. The effect is then likely due to the dye chemistry, not external factors. --Naomi At 03:30 AM 11/20/2005, Ron Ophir wrote: >Hi, >In Chapter 8.1.2 in Limma users guide there is a description how to >detecting the Dye effect. >The example there describes an experiment of Wt vs Mut with two dye >swaps replicates as follow: > >FileName Cy3 Cy5 >File1 wt mu >File2 mu wt >File3 wt mu >File4 mu wt > >Let's say that I found a list of gene which are significant due to >mutant effect and due to dye effect as well. Can I only ignore them or >can I correct the dye effect? >I guess that it depends how the dye swaps replicates was prepared. If >the dye replicates are technical replicates using the block design is >enough to correct dye effect. If dye replicates are also biological >replicates they night also represent a batch effect. That is all first >replicates was sent in one day and the second replicates sent in another >day, which this difference by itself without dye swap may be a source >for variation. The second option of dye swap preparation may be >corrected by ANCOVA (?). Is possible to do it with LIMMA? I know that a >question about using ANCOVA with LIMMA arose by Naomi Altman but this >discussion was beyond my knowledge. So back to my questions: Is >correcting the dye effect is possible in LIMMA? Is ANCOVA is a solution? >and HOW? > >I hope these questions are in the focus of that mailing list, >Thanks >Ron > >_______________________________________________ >Bioconductor mailing list >Bioconductor at stat.math.ethz.ch >https://stat.ethz.ch/mailman/listinfo/bioconductor 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