Dear Erika, I think avereps simply calculates a mean value - which might be problematic if your expression values are log transformed after normalization... Maybe try to apply avereps before normalization and compare the results to those without averepsing. Bests, Claus-J?rgen Erika Melissari schrieb: > > Dear list, > > does anyone use avereps function by limma package to average probe > copies on microarray? > I used this function and obtained a strange result: my differentially > expressed gene list sensibly change if I use or not this function (I > perform the average before lmFit and after normalization). > I know this function only averages probe copies that are on a > microarray, also if they are in multiple not fixed number. > What is the problem? > Thank you very much > > Erika > > _______________________________________________ > 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

Dear Erika, I've read that too and of course avereps accepts MAList objects, so from that point of view the application of avereps is absolutely correct. However, as far as I remember (I hardly work with limma), the M and A values are somehow log-transformed, and avereps simply calculates an average (at least in an initial check that has not been performed on an MAList). Imagine intensities from replicated spots (for simplicity in log10), say 2 and 4. Obviously the average of these values is 3, but re-converted to linearity, the values 100 and 10000 are NOT averaged to 1000. Maybe that's the point... Bests, Claus-J?rgen Erika Melissari schrieb: > Dear Claus-J?rgen, > > on avereps help I read: > > A new data object is computed in which each probe is represented by > the average of its replicate spots. For an MAList object, the > components M and A are both averaged in this way, as weights and any > matrices found in object$other. > For an MAList object, ID defaults to MA$genes$ID is that exists, > otherwise to rownames(MA$M). > > If x is of mode "character", then the replicate values are assumed to > be equal and the first is taken as the average. > > > > Avereps schould works on an MA object, that is after normalization. If > his object is not an MA list, it works like uniquegenelist function I > think, and does not perform the average. > > Erika > > > ----- Original Message ----- From: "Claus-J?rgen Scholz" > <scholz at="" klin-biochem.uni-wuerzburg.de=""> > To: <erika.melissari at="" bioclinica.unipi.it=""> > Cc: <bioconductor at="" stat.math.ethz.ch=""> > Sent: Tuesday, April 28, 2009 13:19 PM > Subject: Re: [BioC] avereps function > > >> Dear Erika, >> >> I think avereps simply calculates a mean value - which might be >> problematic if your expression values are log transformed after >> normalization... Maybe try to apply avereps before normalization and >> compare the results to those without averepsing. >> >> Bests, >> Claus-J?rgen >> >> Erika Melissari schrieb: >>> >>> Dear list, >>> >>> does anyone use avereps function by limma package to average probe >>> copies on microarray? >>> I used this function and obtained a strange result: my differentially >>> expressed gene list sensibly change if I use or not this function (I >>> perform the average before lmFit and after normalization). >>> I know this function only averages probe copies that are on a >>> microarray, also if they are in multiple not fixed number. >>> What is the problem? >>> Thank you very much >>> >>> Erika >>> >>> _______________________________________________ >>> 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