Hi I'm currently using RMA-preprocessed microarray data to look for differentially expressed genes. I used Limma (lmFit, contrast.fit, eBayes) and topTable (adjust='FDR' = Benjamini&Hochberg I think?) and retained the genes with logFC >2 and adjusted P-value <0.05 (=double filtering I think?), but this appears not to be such a good idea (cfr. article)? Should I redo the analysis now? Is it possible to use the methods explained in the article in R? Thanks for helping Koen Marien Master Student Bioscience Engineering: Cell & Gene Biotechnology Univerity of Ghent (Belgium) http://bene.vub.ac.be/Personal%20Pages/KM.htm ---------------------------------------------------------------------- ------ -------------------- This strategy is bound to be less efficient, though. See a recent article on this subject. <http: www.biomedcentral.com="" 1471-2105="" 10="" 402=""> http://www.biomedcentral.com/1471-2105/10/402 -Christos Christos Hatzis, Ph.D. Nuvera Biosciences, Inc. 400 West Cummings Park, Suite 5350 Woburn, MA 01801 781-938-3844 [[alternative HTML version deleted]]

Hi Marien, You should be safe with Limma. That study concluded that shrinkage based methods that use regularized versions of the standard t statistic should be more efficient than the double filtering method (standard t-test & fold change). Limma uses a regularized or moderated t-statistic for individual gene testing. Selecting genes based on FDR would probably be the most sensible approach for getting to the differential genes. Further filtering by FC would be justifiable if interested in genes differentially expressed in one direction or have a minimum effect magnitude. -Christos Christos Hatzis, Ph.D. Nuvera Biosciences, Inc. 400 West Cummings Park, Suite 5350 Woburn, MA 01801 781-938-3844 -----Original Message----- From: bioconductor-bounces@stat.math.ethz.ch [mailto:bioconductor-bounces at stat.math.ethz.ch] On Behalf Of Koen Marien Sent: Monday, February 01, 2010 8:10 AM To: bioconductor at stat.math.ethz.ch Subject: [BioC] question about lmFit model: double filtering not such a good idea? Hi I'm currently using RMA-preprocessed microarray data to look for differentially expressed genes. I used Limma (lmFit, contrast.fit, eBayes) and topTable (adjust='FDR' = Benjamini&Hochberg I think?) and retained the genes with logFC >2 and adjusted P-value <0.05 (=double filtering I think?), but this appears not to be such a good idea (cfr. article)? Should I redo the analysis now? Is it possible to use the methods explained in the article in R? Thanks for helping Koen Marien Master Student Bioscience Engineering: Cell & Gene Biotechnology Univerity of Ghent (Belgium) http://bene.vub.ac.be/Personal%20Pages/KM.htm ---------------------------------------------------------------------- ------ -------------------- This strategy is bound to be less efficient, though. See a recent article on this subject. <http: www.biomedcentral.com="" 1471-2105="" 10="" 402=""> http://www.biomedcentral.com/1471-2105/10/402 -Christos Christos Hatzis, Ph.D. Nuvera Biosciences, Inc. 400 West Cummings Park, Suite 5350 Woburn, MA 01801 781-938-3844 [[alternative HTML version deleted]] _______________________________________________ Bioconductor mailing list Bioconductor at stat.math.ethz.ch https://stat.ethz.ch/mailman/listinfo/bioconductor Search the archives: http://news.gmane.org/gmane.science.biology.informatics.conductor