Hi Richard Vince is spot on. Do you have a reason to worry that the differences in results are non-negligible? Best wishes Wolfgang Vincent Carey scripsit 02/10/2012 01:44 PM: > I believe you need to have a look at the documentation of vsn2 > > subsample: Integer of length 1. If its value is greater than 0, the > model parameters are estimated from a subsample of the data > of size 'subsample' only, yet the fitted transformation is > then applied to all data. For large datasets, this can > substantially reduce the CPU time and memory consumption at a > negligible loss of precision. Note that the 'AffyBatch' > method of 'vsn2' sets a value of '30000' for this parameter > if it is missing from the function call - which is different > from the behaviour of the other methods. > > > On Fri, Feb 10, 2012 at 5:22 AM, Richard Coulson [guest]< > guest at bioconductor.org> wrote: > >> >> Hi, >> >> I have a problem with using 'limma' when I'm analysing some microarray >> data. If I run the below code WITHOUT setting a seed, I get slightly >> different values for the coefficients each time it's run; however this >> problem does not occur if I do set one (e.g. set.seed(1223762671)) :- >> >> raw.data<-ReadAffy( celfile.path="CEL directory" ) >> normalised.data<-vsnrma(raw.data) >> >> transfect.lmFit<-lmFit( normalised.data, design.matrix ) >> cont.lmFit<-contrasts.fit(transfect.lmFit, cont.matrix) >> >> i.e. the values in cont.lmFit$coefficients are altered from one R session >> to another. >> >> Please could anyone help with this? >> >> Many thanks, >> Richard. >> >> >> -- output of sessionInfo(): >> >>> sessionInfo() >> R version 2.12.2 (2011-02-25) >> Platform: x86_64-pc-linux-gnu (64-bit) >> >> locale: >> [1] LC_CTYPE=en_GB.UTF-8 LC_NUMERIC=C >> [3] LC_TIME=en_GB.UTF-8 LC_COLLATE=en_GB.UTF-8 >> [5] LC_MONETARY=C LC_MESSAGES=en_GB.UTF-8 >> [7] LC_PAPER=en_GB.UTF-8 LC_NAME=C >> [9] LC_ADDRESS=C LC_TELEPHONE=C >> [11] LC_MEASUREMENT=en_GB.UTF-8 LC_IDENTIFICATION=C >> >> attached base packages: >> [1] stats graphics grDevices utils datasets methods base >> >> other attached packages: >> [1] hugene11stv1cdf_1.26.0 limma_3.4.4 vsn_3.16.0 >> [4] affyPLM_1.24.0 preprocessCore_1.10.0 gcrma_2.20.0 >> [7] affy_1.26.1 Biobase_2.8.0 >> >> loaded via a namespace (and not attached): >> [1] affyio_1.16.0 Biostrings_2.16.9 grid_2.12.2 IRanges_1.6.8 >> [5] lattice_0.19-17 splines_2.12.2 tools_2.12.2 >> >>> cont.matrix >> Contrasts >> Levels (C.GFP.24+C.GFP.48+C.GFP.72)-(mock.24+mock.48+mock.72) >> C.GFP.24 1 >> C.GFP.48 1 >> C.GFP.72 1 >> mock.24 -1 >> mock.48 -1 >> mock.72 -1 >> myc.24 0 >> myc.48 0 >> myc.72 0 >> N.GFP.24 0 >> N.GFP.48 0 >> N.GFP.72 0 >> untransfected.0 0 >> Contrasts >> Levels (N.GFP.24+N.GFP.48+N.GFP.72)-(mock.24+mock.48+mock.72) >> C.GFP.24 0 >> C.GFP.48 0 >> C.GFP.72 0 >> mock.24 -1 >> mock.48 -1 >> mock.72 -1 >> myc.24 0 >> myc.48 0 >> myc.72 0 >> N.GFP.24 1 >> N.GFP.48 1 >> N.GFP.72 1 >> untransfected.0 0 >> Contrasts >> Levels (myc.24+myc.48+myc.72)-(mock.24+mock.48+mock.72) >> C.GFP.24 0 >> C.GFP.48 0 >> C.GFP.72 0 >> mock.24 -1 >> mock.48 -1 >> mock.72 -1 >> myc.24 1 >> myc.48 1 >> myc.72 1 >> N.GFP.24 0 >> N.GFP.48 0 >> N.GFP.72 0 >> untransfected.0 0 >> >> >> -- >> Sent via the guest posting facility at bioconductor.org. >> >> _______________________________________________ >> Bioconductor mailing list >> Bioconductor at r-project.org >> https://stat.ethz.ch/mailman/listinfo/bioconductor >> Search the archives: >> http://news.gmane.org/gmane.science.biology.informatics.conductor >> > > [[alternative HTML version deleted]] > > _______________________________________________ > Bioconductor mailing list > Bioconductor at r-project.org > https://stat.ethz.ch/mailman/listinfo/bioconductor > Search the archives: http://news.gmane.org/gmane.science.biology.informatics.conductor -- Best wishes Wolfgang Wolfgang Huber EMBL http://www.embl.de/research/units/genome_biology/huber

Hi, Many thanks - setting subsample to 0 seems to have fixed the problem, though it does slow down the normalisation step a bit. Thanks again, Richard. On 10/02/12 12:44, Vincent Carey wrote: > I believe you need to have a look at the documentation of vsn2 > > subsample: Integer of length 1. If its value is greater than 0, the > model parameters are estimated from a subsample of the data > of size 'subsample' only, yet the fitted transformation is > then applied to all data. For large datasets, this can > substantially reduce the CPU time and memory consumption at a > negligible loss of precision. Note that the 'AffyBatch' > method of 'vsn2' sets a value of '30000' for this parameter > if it is missing from the function call - which is different > from the behaviour of the other methods. > > > On Fri, Feb 10, 2012 at 5:22 AM, Richard Coulson [guest]< > guest at bioconductor.org> wrote: > >> Hi, >> >> I have a problem with using 'limma' when I'm analysing some microarray >> data. If I run the below code WITHOUT setting a seed, I get slightly >> different values for the coefficients each time it's run; however this >> problem does not occur if I do set one (e.g. set.seed(1223762671)) :- >> >> raw.data<-ReadAffy( celfile.path="CEL directory" ) >> normalised.data<-vsnrma(raw.data) >> >> transfect.lmFit<-lmFit( normalised.data, design.matrix ) >> cont.lmFit<-contrasts.fit(transfect.lmFit, cont.matrix) >> >> i.e. the values in cont.lmFit$coefficients are altered from one R session >> to another. >> >> Please could anyone help with this? >> >> Many thanks, >> Richard. >> >> >> -- output of sessionInfo(): >> >>> sessionInfo() >> R version 2.12.2 (2011-02-25) >> Platform: x86_64-pc-linux-gnu (64-bit) >> >> locale: >> [1] LC_CTYPE=en_GB.UTF-8 LC_NUMERIC=C >> [3] LC_TIME=en_GB.UTF-8 LC_COLLATE=en_GB.UTF-8 >> [5] LC_MONETARY=C LC_MESSAGES=en_GB.UTF-8 >> [7] LC_PAPER=en_GB.UTF-8 LC_NAME=C >> [9] LC_ADDRESS=C LC_TELEPHONE=C >> [11] LC_MEASUREMENT=en_GB.UTF-8 LC_IDENTIFICATION=C >> >> attached base packages: >> [1] stats graphics grDevices utils datasets methods base >> >> other attached packages: >> [1] hugene11stv1cdf_1.26.0 limma_3.4.4 vsn_3.16.0 >> [4] affyPLM_1.24.0 preprocessCore_1.10.0 gcrma_2.20.0 >> [7] affy_1.26.1 Biobase_2.8.0 >> >> loaded via a namespace (and not attached): >> [1] affyio_1.16.0 Biostrings_2.16.9 grid_2.12.2 IRanges_1.6.8 >> [5] lattice_0.19-17 splines_2.12.2 tools_2.12.2 >> >>> cont.matrix >> Contrasts >> Levels (C.GFP.24+C.GFP.48+C.GFP.72)-(mock.24+mock.48+mock.72) >> C.GFP.24 1 >> C.GFP.48 1 >> C.GFP.72 1 >> mock.24 -1 >> mock.48 -1 >> mock.72 -1 >> myc.24 0 >> myc.48 0 >> myc.72 0 >> N.GFP.24 0 >> N.GFP.48 0 >> N.GFP.72 0 >> untransfected.0 0 >> Contrasts >> Levels (N.GFP.24+N.GFP.48+N.GFP.72)-(mock.24+mock.48+mock.72) >> C.GFP.24 0 >> C.GFP.48 0 >> C.GFP.72 0 >> mock.24 -1 >> mock.48 -1 >> mock.72 -1 >> myc.24 0 >> myc.48 0 >> myc.72 0 >> N.GFP.24 1 >> N.GFP.48 1 >> N.GFP.72 1 >> untransfected.0 0 >> Contrasts >> Levels (myc.24+myc.48+myc.72)-(mock.24+mock.48+mock.72) >> C.GFP.24 0 >> C.GFP.48 0 >> C.GFP.72 0 >> mock.24 -1 >> mock.48 -1 >> mock.72 -1 >> myc.24 1 >> myc.48 1 >> myc.72 1 >> N.GFP.24 0 >> N.GFP.48 0 >> N.GFP.72 0 >> untransfected.0 0 >> >> >> -- >> Sent via the guest posting facility at bioconductor.org. >> >> _______________________________________________ >> Bioconductor mailing list >> Bioconductor at r-project.org >> https://stat.ethz.ch/mailman/listinfo/bioconductor >> Search the archives: >> http://news.gmane.org/gmane.science.biology.informatics.conductor >>