Rich, Unless I am mistaken, it doesn't sound like you have made a CDF environment for your custom probe sets yet. You will have to do that first. You need to use makecdfenv to create a package containing the environment, then install the package. Supposing your CDF package is called newcdf, you could then do Data <- ReadAffy() Data@cdfName <- "new" # I think you have to omit the "cdf" part As for using gcrma, you'll have to make a separate probe package for this. IIRC Wolfgang Huber posted a way to do that fairly recently, using matchprobes - try the mailing list archive. Cheers, Ken > I wrote a program that takes an original text CDF > file as a template and writes out my revised "units" > (Affyspeak for probesets). The revised units are sets > of probes that map to a quality filtered set of > transcripts: higher quality transcripts, chimerics > stripped, latest Genbank records, etc. A unit > (probeset) can get very large and I'm going to try and > reuse probes if it's possible (sometimes probes > match to mulitple genes). > > I'm hoping that this will work with RMA but I am > concerned about GCRMA since it relies on additional > probeset sequence information which is originally > available in the "probetab" files provided by > Affymetrix. > > Another possible concern is if Robert Genetlemen is > right and I have to change the chipname in the CEL > file. This is, of course, a piece of cake with TEXT > CEL files. Binary CEL files will require a little > more delicate approach: writing a quick hack program > to replace the old string. > > I will certainly check out the altcdfenv package. > > Again, if anyone has any caveats when dealing with > this, pleas let me and everybody else know about it. > > Rich >