In LPE, rank-invariant resampling based FDR is used. For manuscript of the paper, please visist http://www.healthsystem.virginia.edu/internet/hes/biostat/bioinformati cs/ z-scores need not follow N(0,1) distribution. -N ______________________ Nitin Jain, PhD Non Clinical Statistics Pfizer, Inc. (Groton, CT) -----Original Message----- From: bioconductor-bounces@stat.math.ethz.ch [mailto:bioconductor-bounces@stat.math.ethz.ch]On Behalf Of Harry Hurd Sent: Tuesday, February 22, 2005 10:20 PM To: bioconductor@stat.math.ethz.ch Subject: [BioC] LPE Hello, We have recently been trying LPE as implemented in bioconductor. We tried it for N=3 on the affy 133plus array (55K probes). The most significant results are very pleasing to the biologists (so far anyway) but in trying to do some further work to assign some FDRs, I plotted the CDF and a normal probability plot for the resulting z scores. There is a huge proportion of small values. The zscores do not look anything like normal. To what extent should the z scores look N(0,1)? The estimated sigma is about .5; should it be near 1? Does anyone have experience using LPE on a large array like the 133plus? Thanks much for any help. Harry Hurd Dept of Statistics UNC Chapel Hill [[alternative HTML version deleted]] _______________________________________________ Bioconductor mailing list Bioconductor@stat.math.ethz.ch https://stat.ethz.ch/mailman/listinfo/bioconductor LEGAL NOTICE\ Unless expressly stated otherwise, this messag...{{dropped}}