Hi Wolfgang, Actually there is an as.matrix method for vsn objects in the limma package, it just had a bug. Now fixed. Best wishes Gordon >[BioC] Problem with "as.matrix.vsn" and "duplicateCorrelation" fromlimma > > > >Wolfgang Huber ><mailto:bioconductor%40stat.math.ethz.ch?subject=%5bbioc%5d%20problem %20with%20%22as.matrix.vsn%22%20and%20%22duplicatecorrelation%22%0a%20="" from%09limma&in-reply-="" to="3aa934e00709151315m371c208dleea83bf372da2aaa%40mail.gmail.com">huber >at ebi.ac.uk >Sun Sep 16 02:07:38 CEST 2007 > >Dear Artur, > >there is no "as.matrix" function for objects of type "vsn", such as >returned by the "vsn2" function. You are right to use the "exprs" >accessor instead. > >If you are looking for a wrapper to vsn2 that produces an ExpressionSet >(rather than a vsn object), please look at "justvsn" (in the devel >branch, will be in the bioc 2.1 release). > >Best wishes > Wolfgang > > > Hi all, > > > > I just wanted to report an error that I've been geting: > > When I run duplicateCorrelation on this vsn object I get an error message. > > If instead of using the vsn object I use the result of an "expres( > > vsn.object)" function call it apparently works. > > I could be mistaken, but I don't think that I upgraded my packages or R and > > duplicateCorrelation after the last time I used duplicateCorrelation > > succesfully. > > > > Thanks a lot! > > > > Artur Veloso > > > > Masters in Marine Biology Candidate > > College of Charleston, South Carolina, USA

Dear Gordon, great - thanks for pointing this out! Just to explain the rationale for "vsn" objects: "vsn" objects are returned by the function "vsn2" and contain all the information on the fit of the vsn model to data. There is a "predict" method that lets users apply the fitted model to either the same or new data (e.g. if you use known housekeeping or spike-in transcrips for fitting and then apply to all data). In many use cases, one just wants to use the normalized & transformed data and not bother with the details of vsn. The wrapper "justvsn" is for that purpose, it returns an ExpressionSet or NChannelSet, as appropriate. In vsn version 1, I had tried to mush the fit results and diagnostics into the returned exprSet object, but that never really worked well, hence it is now done as described above. Best wishes Wolfgang ------------------------------------------------------------------ Wolfgang Huber EBI/EMBL Cambridge UK http://www.ebi.ac.uk/huber > > > Hi Wolfgang, > > Actually there is an as.matrix method for vsn objects in the limma > package, it just had a bug. Now fixed. > > Best wishes > Gordon > >> [BioC] Problem with "as.matrix.vsn" and "duplicateCorrelation" fromlimma >> >> >> >> Wolfgang Huber >> <mailto:bioconductor%40stat.math.ethz.ch?subject=%5bbioc%5d%20probl em%20with%20%22as.matrix.vsn%22%20and%20%22duplicatecorrelation%22%0a%="" 20from%09limma&in-reply-="" to="3aa934e00709151315m371c208dleea83bf372da2aaa%40mail.gmail.com">huber >> at ebi.ac.uk >> Sun Sep 16 02:07:38 CEST 2007 >> >> Dear Artur, >> >> there is no "as.matrix" function for objects of type "vsn", such as >> returned by the "vsn2" function. You are right to use the "exprs" >> accessor instead. >> >> If you are looking for a wrapper to vsn2 that produces an ExpressionSet >> (rather than a vsn object), please look at "justvsn" (in the devel >> branch, will be in the bioc 2.1 release). >> >> Best wishes >> Wolfgang >> >> > Hi all, >> > >> > I just wanted to report an error that I've been geting: >> > When I run duplicateCorrelation on this vsn object I get an error >> message. >> > If instead of using the vsn object I use the result of an "expres( >> > vsn.object)" function call it apparently works. >> > I could be mistaken, but I don't think that I upgraded my packages >> or R and >> > duplicateCorrelation after the last time I used duplicateCorrelation >> > succesfully. >> > >> > Thanks a lot! >> > >> > Artur Veloso >> > >> > Masters in Marine Biology Candidate >> > College of Charleston, South Carolina, USA