Hello, I am new to bioconductor and I have a question about fitting and normalizing the spike-in hybridization controls. >From what I have read so far, it seems that intensity and concentration can be related by way of a langmuir isotherm. The concentrations of the hybridization controls for Affymetrix GeneChips is a known quantity. Does anyone here fit the the intensities of the hybridization controls and then use the fit data to normalize between replicates - like a scale factor? If so, do you have a procedure? If not, is there a reason that one should not do this? Thanks Matt

Hi Matt, On 11/8/2012 6:57 PM, Thornton, Matthew wrote: > Hello, > > I am new to bioconductor and I have a question about fitting and normalizing the spike-in hybridization controls. > > > From what I have read so far, it seems that intensity and concentration can be related by way of a langmuir isotherm. The concentrations of the hybridization controls for Affymetrix GeneChips is a known quantity. Does anyone here fit the the intensities of the hybridization controls and then use the fit data to normalize between replicates - like a scale factor? It wouldn't be a scale factor - that would imply a simple shift of the distribution, for which you don't need to fit a model. There is a long history of people using various forms of control spots of different concentrations to fit a linear model of some type and using parameters of that model to normalize data. You can use known spike ins like you suggest, or you can use Li and Wong's idea of finding invariant probesets and fitting a loess model. There are several other possibilities as well. So it isn't unheard of to do what you suggest. However, this is ground that was worked over fairly thoroughly in the middle of the last decade, and except in fairly pathological cases (e.g., when a preponderance of genes are differentially expressed), a simple quantile normalization seems to work pretty well, and has since become more or less the paradigm. Best, Jim > > If so, do you have a procedure? If not, is there a reason that one should not do this? > > Thanks > > Matt > _______________________________________________ > 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 -- James W. MacDonald, M.S. Biostatistician University of Washington Environmental and Occupational Health Sciences 4225 Roosevelt Way NE, # 100 Seattle WA 98105-6099