All, Is the MA plot only useful as a visualization diagnostic under the assumption that most genes are not changing? Since we compare to a reference array, it seems that if most genes change, then the creation of this reference could be problematic. Does anyone know if any literature that addresses this. What normalization diagnostics may one use under the assumption that many genes are changing. Thanks, Juliet [[alternative HTML version deleted]]

On Apr 24, 2013, at 9:43 PM, Juliet Hannah <juliet.hannah at="" gmail.com=""> wrote: > All, > > Is the MA plot only useful as a visualization diagnostic under the > assumption that most genes are not changing? Since we compare to a > reference array, it seems that if most genes change, then the creation of > this reference could be problematic. Does anyone know if any literature > that addresses this. What normalization diagnostics may one use under the > assumption that many genes are changing. > Dear Juliet: The answer can take two directions: 1. Use a control set of genes for which you have either previous knowledge that they are not changing, or for which you can establish that they are not changing by an independent method. One can argue that every well-designed experiment builds upon some prior knowledge and/or independent validation technology. 2. Decompose the changes into a common main trend that holds for "most" genes, and then the differences from that. The first component of this decomposition is usually absorbed by "normalisation" methods such as vsn, loess, quantiles, but can be reconstructed from their parameters. Best wishes Wolfgang > Thanks, > > Juliet > > [[alternative HTML version deleted]] > > _______________________________________________ > Bioconductor mailing list > Bioconductor at r-project.org > https://stat.ethz.ch/mailman/listinfo/bioconductor > Search the archives: http://news.gmane.org/gmane.science.biology.informatics.conductor