Hello. I am examining data from a 2D gel experiment, but my question is about a broader issue of normalization. Each gel has been stained with two different dyes, whose respective intensities report two different protein features, let's call them A and B. The quantity of interest is A/B. The data consists of two protein-by-gel matrices of intensities, and I need to normalize them. The questions is, which of the following would be the preferrable choice? 1) First do normalize.quantiles() on the A and B matrices individually and then calculate A/B. 2) First calculate A/B and then do normalize.quantiles() on that. Thank you so much!

Alex Bokov ha scritto: > Hello. I am examining data from a 2D gel experiment, but my question is > about a broader issue of normalization. > > Each gel has been stained with two different dyes, whose respective > intensities report two different protein features, let's call them A and > B. The quantity of interest is A/B. The data consists of two > protein-by-gel matrices of intensities, and I need to normalize them. > The questions is, which of the following would be the preferrable choice? > > 1) First do normalize.quantiles() on the A and B matrices individually > and then calculate A/B. > 2) First calculate A/B and then do normalize.quantiles() on that. > > Thank you so much! > Dear Alex have you checked whether it actually makes a difference - and if so, in what way? Best wishes Wolfgang ---------------------------------------------------- Wolfgang Huber, EMBL-EBI, http://www.ebi.ac.uk/huber