Dear Bioconductor users, Before I performed Limma analysis, should I do non specific filter or not? If we should do filter before Limma step, how do I determine the cut- off for filtering-function? For example, If I use Co-efficient of Variability (CV) as filtering criteria, how should I set the cut-off? arbitrarily? Thanks in advance. Yisong [[alternative HTML version deleted]]

Hi Yisong, One of the assumptions behind the eBayes() step used by limma is that you have not filtered out any genes (e.g., the eBayes() step estimates the 'average' variability of the genes on your microarray, and if you filter some out, you can affect this estimate). As for how to filter, this has been the subject of many many conversations on this list, and I think there are as many opinions as subscribers ;-D. IMO, whatever you do you should have a reasonable rationale for what you have done, which is true for most steps in a given analysis. Since I can't give you a rationale I won't belabor you with my opinion either. Best, Jim James W. MacDonald, M.S. Biostatistician Hildebrandt Lab 8220D MSRB III 1150 W. Medical Center Drive Ann Arbor MI 48109-0646 734-936-8662 >>> "Yisong Zhen" <zhenyisong at="" gmail.com=""> 12/08/08 5:25 PM >>> Dear Bioconductor users, Before I performed Limma analysis, should I do non specific filter or not? If we should do filter before Limma step, how do I determine the cut- off for filtering-function? For example, If I use Co-efficient of Variability (CV) as filtering criteria, how should I set the cut-off? arbitrarily? Thanks in advance. Yisong [[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 ********************************************************** Electronic Mail is not secure, may not be read every day, and should not be used for urgent or sensitive issues

Hi Yisong, You want to filter AFTER doing the eBayes() correction, but BEFORE doing the multiple test correction. This way, eBayes() correctly estimates the "average" variability of the genes, but you only have to do the multiple test correction on the genes for which you can rationally create a filter. Cheers, Jenny At 05:26 PM 12/8/2008, Yisong Zhen wrote: >It seems that I have to read carefully the Limma module. I though it will >reduce the false positive ( and following step about multiple- testing >correction) when filtered out those genes with low expression constantly and >genes with low CV. >Yisong >On Mon, Dec 8, 2008 at 3:48 PM, James MacDonald <jmacdon at="" med.umich.edu="">wrote: > > > Hi Yisong, > > > > One of the assumptions behind the eBayes() step used by limma is that you > > have not filtered out any genes (e.g., the eBayes() step estimates the > > 'average' variability of the genes on your microarray, and if you filter > > some out, you can affect this estimate). > > > > As for how to filter, this has been the subject of many many conversations > > on this list, and I think there are as many opinions as subscribers ;-D. > > IMO, whatever you do you should have a reasonable rationale for what you > > have done, which is true for most steps in a given analysis. Since I can't > > give you a rationale I won't belabor you with my opinion either. > > > > Best, > > > > Jim > > > > > > > > James W. MacDonald, M.S. > > Biostatistician > > Hildebrandt Lab > > 8220D MSRB III > > 1150 W. Medical Center Drive > > Ann Arbor MI 48109-0646 > > 734-936-8662 > > >>> "Yisong Zhen" <zhenyisong at="" gmail.com=""> 12/08/08 5:25 PM >>> > > Dear Bioconductor users, > > > > Before I performed Limma analysis, should I do non specific filter or not? > > If we should do filter before Limma step, how do I determine the cut-off > > for > > filtering-function? For example, If I use Co-efficient of Variability (CV) > > as filtering criteria, how should I set the cut-off? arbitrarily? > > > > Thanks in advance. > > > > Yisong > > > > [[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 > > > > ********************************************************** > > Electronic Mail is not secure, may not be read every day, and should not be > > used for urgent or sensitive issues > > > > > >-- >//------------------------------------------------------------------- ---------------------- >// We have a hunger of the mind which asks for knowledge >// of all around us, and the more we gain, the more is >// our desire; the more we see, the more we are capable >// of seeing. >//------------------------------------------------------------------- --------------------- > >Yisong Zhen, Ph.D (Davidson Lab) >Molecular & Cellular Biology >Molecular Cardiovascular Research Program >University of Arizona >1656 E. Mabel, MRB 317 >Tucson, AZ 85724 USA >Lab: 520-626-8153 >Cell: 520-977-7397 > >@-------------------------------------------------------------------- ------------------- > > [[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 Jenny Drnevich, Ph.D. Functional Genomics Bioinformatics Specialist W.M. Keck Center for Comparative and Functional Genomics Roy J. Carver Biotechnology Center University of Illinois, Urbana-Champaign 330 ERML 1201 W. Gregory Dr. Urbana, IL 61801 USA ph: 217-244-7355 fax: 217-265-5066 e-mail: drnevich at illinois.edu