Thanks for the info, guys. Abhi, nice to hear from you again! The vignette does show some intriguing information that I'll have to work through. We do have replicates in this particular experiment so there might the way I'm running (which is what page 6 is suggesting to me). I'll post again once I work through some data. todd From: Vince S. Buffalo [mailto:vsbuffalo@gmail.com] Sent: Monday, March 12, 2012 7:28 PM To: Abhishek Pratap Cc: Creasy, Todd; bioconductor@r-project.org Subject: Re: [BioC] Newer version of DESeq give much small list of DE genes The new version of DESeq is more conservative - we've seen this too. As far as I understand, this is because of a change in the way it shares information between per-gene empirical variance estimates and the fitted variance (which is parametric and no longer local in the new version). There is some good information on the bottom of page 6 of the vignette: http://www.bioconductor.org/packages/release/bioc/vig nettes/DESeq/inst/doc/DESeq.pdf HTH, Vince On Mon, Mar 12, 2012 at 3:50 PM, Abhishek Pratap <apratap@lbl.gov<mailto:apratap@lbl.gov>> wrote: Hi Todd I have seen this behaviour when I reduce the number of replicates to 0. I am assuming in your case with both the DESeq versions you are using same number of replicates. Also in case you are not using replicates at all then I believe it can affect the latest .DESeq model. I am not sure about the version 1.2. -Abhi On Mon, Mar 12, 2012 at 3:11 PM, Creasy, Todd <tcreasy@som.umaryland.edu<mailto:tcreasy@som.umaryland.edu>> wrote: > Hi, > > We're comparing an older version of DESeq with the new 1.6 version and I'm finding a large discrepancy. Probably an option I'm missing or something. In the samples I'm working with, I see total DE genes going from ~200 - 2000 genes at FDR = 0.10 to nearly 0 genes using the new version at the same FDR. > > What could be going on here. Admittedly, the old version was 1.2 so we were way behind but I'm still shocked by this drop off. > > Thoughts? > > todd > > [[alternative HTML version deleted]] > > _______________________________________________ > Bioconductor mailing list > Bioconductor@r-project.org<mailto:bioconductor@r-project.org> > https://stat.ethz.ch/mailman/listinfo/bioconductor > Search the archives: http://news.gmane.org/gmane.science.biology.informatics.conductor _______________________________________________ Bioconductor mailing list Bioconductor@r-project.org<mailto:bioconductor@r-project.org> https://stat.ethz.ch/mailman/listinfo/bioconductor Search the archives: http://news.gmane.org/gmane.science.biology.informatics.conductor -- Vince Buffalo Statistical Programmer Bioinformatics Core UC Davis Genome Center vincebuffalo.com<http: vincebuffalo.com=""/> twitter.com/vsbuffalo<http: twitter.com="" vsbuffalo=""> [[alternative HTML version deleted]]