Dear All, I am currently using an Agilent 4x44K model organism mosquito array for a one color time-course course experiment. I am analyzing using the LIMMA package and have performed background correction and between array normalization as follows. RoBb.corr <- backgroundCorrect(RoB, method="normexp", offset=16) RoBb.corr.norm <- normalizeBetweenArrays(RoBb.corr, method="quantile") However, after background correction the boxplot for a number of arrays become compressed (see example here: https://dl.dropbox.com/u/407047/Work/Catteruccia/minExample.html ). I am not sure what is causing this compression although the quantile between array normalisation seems to correct for this . However I am concerned about the possible affect on the data. Has anyone else seen this compression with normexp correction? I have read that background correction is not always optimal for Agilent arrays (Zahurak et al. 2007 BMC Bioinformatics). Do others routinely omit the background correction for Agilent arrays? Best regards Dr Sara Mitchell Imperial College London [[alternative HTML version deleted]]

> Dear All, > Hi Sara, > I am currently using an Agilent 4x44K model organism mosquito array for a > one color time-course course experiment. I am analyzing using the LIMMA > package and have performed background correction and between array > normalization as follows. > > RoBb.corr <- backgroundCorrect(RoB, method="normexp", offset=16) > > RoBb.corr.norm <- normalizeBetweenArrays(RoBb.corr, method="quantile") > > However, after background correction the boxplot for a number of arrays > become compressed (see example here: > https://dl.dropbox.com/u/407047/Work/Catteruccia/minExample.html ). > With such an extreme correction, it looks like it might be your arrays that are the problem, rather than the specific background correction method. Have you tried producing similar boxplots just for the background values? Or even looking at the actual images from the scan. If some of the arrays have a uniformly high signal for both foreground and background values, it could indicate that the hybridisation somehow failed (maybe too much salt in the sample, which causes an all-over high intensity?). Apart from that, yes I have tried using Agilent arrays without any background correction methods. But I definitely wouldn't recommend it in this case, until you figure out what's going on with those outlier arrays. And I would NOT recommend just using quantile normalisation to just make the distributions on all arrays similar when they're so highly different to begin with. HTH \Heidi > I am not sure what is causing this compression although the quantile > between array normalisation seems to correct for this . However I am > concerned about the possible affect on the data. Has anyone else seen > this compression with normexp correction? > > I have read that background correction is not always optimal for Agilent > arrays (Zahurak et al. 2007 BMC Bioinformatics). > > > Do others routinely omit the background correction for Agilent arrays? > > Best regards > > > Dr Sara Mitchell > > Imperial College London > > [[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