Sorry, my intent was to return it to the list. Needing devel explains why a cutoff of 0.5 only returned 674 probe sets with nsFilter. That had me scratching my head plenty but after the brain fade with rowQ, I wasn't going to ask any questions until I was more sure of my footing with the genefilter package. Dennis Burian, Ph.D. Functional Genomics Group Civil Aerospace Medical Institute, AAM-610 6500 S. MacArthur Blvd. Oklahoma City OK 73169 405-954-6087 dennis.burian at faa.gov Robert Gentleman <rgentlem at="" fhcrc.o="" rg=""> To Dennis M Burian/AMC/FAA at FAA 01/31/2008 05:23 cc PM Subject Re: [BioC] rowQ error Yes, then the right thing is to try out nsFilter, and if it works (or doesn't), if you could return the thread to the mailing list, then all will see its successful conclusion, (oh and you need to use the devel version as nsFilter is broken in release). otherwise something like lowQ = rowQ(myexprset, floor(0.25*numberofsamples)) upQ = rowQ(myexprset, ceiling(0.75*numberofsamples)) iqr = upQ-lowQ good = iqr > median(iqr) newEset = myexprset[good,] thanks Robert Dennis.Burian at faa.gov wrote: > First, thanks for the quick reply and I'll look at nsFilter. Per our > discussion at the Introductory R course last November, the intent here is > to keep the top 50% of genes based on IQR for differential expression > testing. I'm following the script in the Machine Learning module from the > course but I'm open to suggestion. > > > Dennis Burian, Ph.D. > Functional Genomics Group > Civil Aerospace Medical Institute, AAM-610 > 6500 S. MacArthur Blvd. > Oklahoma City OK 73169 > 405-954-6087 > dennis.burian at faa.gov > -- Robert Gentleman, PhD Program in Computational Biology Division of Public Health Sciences Fred Hutchinson Cancer Research Center 1100 Fairview Ave. N, M2-B876 PO Box 19024 Seattle, Washington 98109-1024 206-667-7700 rgentlem at fhcrc.org