Hi Bioconductor User; There's a new package on in the Developmental Packages section of www.bioconductor.org called RankProd. It identifies differentially expressed genes and estimates the associated percentage of false prediction (also known as false discovery rate) using the Rank Product method of Breitling, et al., FEBS Letters 573 (2004) 83-92 It is a robust nonparametric method, which would gain power in the case of low number of replications and/or relatively large scale of noise. The method can natural combine data sets from similar experiments conducted under different conditions, e.g., different laboratories, different experimenters, treatment with different concentrations... (Please refer the package vignette for detail) Any comment or suggestion is welcome. Fangxin -- Fangxin Hong, Ph.D. Plant Biology Laboratory The Salk Institute 10010 N. Torrey Pines Rd. La Jolla, CA 92037 E-mail: fhong@salk.edu