Hi, I am using limma to analyse an experiment where I am comparing the response in stimulated verses un-stimulated cells in individuals with and without disease. When I ask how individuals with or without disease respond differently to the stimulus there are no significant genes when the p values are adjusted. I know that there are differences which have been confirmed by qRT-PCR ( and can be demonstarted by analysing data using fold change only) and these genes have the highest ranked p values in the limma analysis (although not significant when adjusted). I have tried to filter the data set (to the 3000 most variable genes) so there are less comparisons being made and the differences are still not significant. I am using hgu133plus2 chips with 3 replicates. regards Anthony -- ______________________________________________ Anthony Bosco - PhD Student Institute for Child Health Research (Company Limited by Guarantee ACN 009 278 755) Subiaco, Western Australia, 6008 Ph 61 8 9489 , Fax 61 8 9489 7700 email anthonyb@ichr.uwa.edu.au

Although some genes have significant differences, it is not necesssarily for limma to identify them out. In other words, genes that are not identified by limma may be significant. It all depends on the power of your test, which may in turn decided by your sample size, your error varaince (noise scale) and the test you choose. Sometimes, some significant differences might be small for your data to identify them out. Since you have done the experiment, you can't change sample size and error variance, maybe you can try other method or low down the overall sifnificance you controlled. Hopefully this helps. Bests; Fangxin > Hi, > > I am using limma to analyse an experiment where I am comparing the > response in stimulated verses un-stimulated cells in individuals with > and without disease. > > When I ask how individuals with or without disease respond > differently to the stimulus there are no significant genes when the p > values are adjusted. > > I know that there are differences which have been confirmed by > qRT-PCR ( and can be demonstarted by analysing data using fold change > only) and these genes have the highest ranked p values in the limma > analysis (although not significant when adjusted). > > I have tried to filter the data set (to the 3000 most variable genes) > so there are less comparisons being made and the differences are > still not significant. > > I am using hgu133plus2 chips with 3 replicates. > > > regards > > > Anthony > -- > ______________________________________________ > > Anthony Bosco - PhD Student > > Institute for Child Health Research > (Company Limited by Guarantee ACN 009 278 755) > Subiaco, Western Australia, 6008 > > Ph 61 8 9489 , Fax 61 8 9489 7700 > email anthonyb@ichr.uwa.edu.au > > _______________________________________________ > Bioconductor mailing list > Bioconductor@stat.math.ethz.ch > https://stat.ethz.ch/mailman/listinfo/bioconductor > > -- Fangxin Hong, Ph.D. Plant Biology Laboratory The Salk Institute 10010 N. Torrey Pines Rd. La Jolla, CA 92037 E-mail: fhong@salk.edu

>Date: Fri, 5 Nov 2004 03:31:08 +0100 >From: Anthony Bosco <anthonyb@ichr.uwa.edu.au> >Subject: [BioC] limma power question >To: bioconductor@stat.math.ethz.ch >Message-ID: <f05111a06bdb094e4101c@[10.0.5.134]> >Content-Type: text/plain; charset="us-ascii" ; format="flowed" > >Hi, > >I am using limma to analyse an experiment where I am comparing the >response in stimulated verses un-stimulated cells in individuals with >and without disease. > >When I ask how individuals with or without disease respond >differently to the stimulus there are no significant genes when the p >values are adjusted. > >I know that there are differences which have been confirmed by >qRT-PCR ( and can be demonstarted by analysing data using fold change >only) and these genes have the highest ranked p values in the limma >analysis (although not significant when adjusted). > >I have tried to filter the data set (to the 3000 most variable genes) >so there are less comparisons being made and the differences are >still not significant. > >I am using hgu133plus2 chips with 3 replicates. What is your question? As far as I know, the limma method has as good or better power than competing methods for this problem, but no guarantee is offered that significant results will always be provided. If you already have a small group of genes that you have an apriori interest in, you can test for these genes only without adjusting the p-values. Gordon >regards > > >Anthony >-- >______________________________________________ > >Anthony Bosco - PhD Student > >Institute for Child Health Research >(Company Limited by Guarantee ACN 009 278 755) >Subiaco, Western Australia, 6008 > >Ph 61 8 9489 , Fax 61 8 9489 7700 >email anthonyb@ichr.uwa.edu.au