Hello, I've a question concerning Gquantile (also Rquantile) normalization between arrays. Did I understand it right that this normalization "prefers" one channel (G or R) without regarding the experimental design, which may be given by a design matrix. So any dye swap is not respected. Is that right? For Tquantile: If I define targets for normalization all the other none defined are not processed? Or did they treated like using the composite normalization method ? Thanks! Sebastian

At 07:32 PM 26/07/2004, S_Thieme wrote: >Hello, > >I've a question concerning Gquantile (also Rquantile) normalization >between arrays. Did I understand it right that this normalization >"prefers" one channel (G or R) without regarding the experimental design, >which may be given by a design matrix. So any dye swap is not respected. >Is that right? Yes. Gquantile normalization is designed for experiments with a common reference always on the green channel. Quantile normalization is applied to the green channels and then the red channel is adjusted proportionally for each spot on each array. >For Tquantile: If I define targets for normalization all the other none >defined are not processed? Tquantile applies quantile normalization separately to subgroups of channels. You should specify a target name for each channel on each array so that everything is normalized. > Or did they treated like using the composite normalization method ? This has nothing to do with "composite" normalization, which is a within-array normalization technique. Gordon >Thanks! > >Sebastian