Hello I am re-analyzing two of studies that have been performed some time ago to compare good (GR) and bad (BR) responders to a treatment. The most recent study has been done using hgu133plus2. The other used a combination of hgu133plus and hgu133plus2 arrays. Specifically, two thirds of the arrays were hibridized on hgu133plus and one third was hibridized with hgu133plus2. I would like to combine first all the arrays in the first study in order to normalize them together and repeat the comparisons that were performed and then I would like to compare it with the newest one. One of my collaborators performed RMA separately on the two sets of chips and then merged them manually retaining only those probesets whose name was the same, but what I would like to do is to normalize all the arrays at once in order to avoid batch effects. I am aware that I can handle thes batch effects by adding the appropriate factor in tha analysis model but I would prefer to deal with this before the analysis starts. I wonder if someone has has had a similar experience and knows "the best" way to proceed in such situation or a tool who can manage the combination of the arrays prior to normalization. Thanks for the advice Alex Sanchez Statistics Department. University of Barcelona. & Statistics and Bioinformatics Unit. IR-HUVH. [[alternative HTML version deleted]]

Hi, Alex, I performed a similar analysis recently, and corrected batch effects using the method described in this article: http://biostatistics.oxfordjournals.org/cgi/content/full/8/1/118 The authors provide an R-script to run the calculations on your matrices. You should be able to find a link for it in the article. Good luck! Charles