Dear all, I've been struggling with a quality control for a set of Affymetrix hgu133a2 chips created from seven different samples (two chips each, one original sample, the second one with a subpopulation of FACS- sorted cells). Lacking experience with array analysis I was hoping someone from the list would be able to identify what the problem might be. I've posted the standard QCReport over here (not sure the list can handle attachments?): http://fiamh.info/hsph/QualityControl.pdf To a newbie like me it _looks_ fairly okay. From RNA degradation plots I already knew that E_neg (chip 9) was a problem and it's been discarded; what seemed odd is the downscaling of all chips (page 3). Also, the probesets for the cell surface marker used in the FACS sorting exhibited a very strong foldchange (between 20-30 times, to be expected) whereas no other probeset showed more than 2-3-fold up- or down-regulation. What confuses me is the arrayQualityMetrics report. The variance mean dependency chart shows quite a bit of variation for the lower ranks: http://fiamh.info/hsph/report_sample/QMreport.html Mini-script is attached below. I was merely wondering whether the QC is still within normal (expected) boundaries, or whether they warrant further investigations. Thanks! -- Oliver ----8<---- library(affyQCReport); library(simpleaffy); library(arrayQualityMetrics) msc.data <- read.affy('covdesc', path='../CEL Files') QCReport(msc.data, file = "QualityControl.pdf"); msc.eset <- call.exprs(msc.data, "mas5"); arrayQualityMetrics(expressionset=msc.eset, outdir='report_sample', force=TRUE, do.logtransform=FALSE) -----8<------ > sessionInfo() R version 2.7.1 (2008-06-23) i386-apple-darwin8.10.1 locale: C/UTF-8/C/C/C/C attached base packages: [1] grid splines tools stats graphics grDevices utils [8] datasets methods base other attached packages: [1] arrayQualityMetrics_1.6.1 beadarray_1.8.0 [3] latticeExtra_0.5-1 vsn_3.6.0 [5] limma_2.14.5 affyQCReport_1.18.0 [7] geneplotter_1.18.0 annotate_1.18.0 [9] AnnotationDbi_1.2.2 RSQLite_0.6-9 [11] DBI_0.2-4 lattice_0.17-8 [13] RColorBrewer_1.0-2 affyPLM_1.16.0 [15] xtable_1.5-2 simpleaffy_2.16.0 [17] gcrma_2.12.1 matchprobes_1.12.0 [19] genefilter_1.20.0 survival_2.34-1 [21] affy_1.18.2 preprocessCore_1.2.0 [23] affyio_1.8.0 Biobase_2.0.1 loaded via a namespace (and not attached): [1] KernSmooth_2.22-22 -- Research Fellow Department of Biostatistics Harvard School of Public Health

Hi Oliver 1. what happens to the variance-mean plot when you remove the bad array(s)? 2. how does it look like when you use RMA instead of MAS5? (and possibly, vsnrma) 3. The behavior of the variance-mean plot in the upper intensity range is indicative of likely saturation effects. 4. One possible explanation for the behavior of the variance-mean plot in the lower intensity range could be that different (batches of) arrays have quite different background intensity distributions. Best wishes Wolfgang ------------------------------------------------------------------ Wolfgang Huber EBI/EMBL Cambridge UK http://www.ebi.ac.uk/huber 28/08/2008 18:53 Oliver Hofmann scripsit > Dear all, > > > I've been struggling with a quality control for a set of Affymetrix > hgu133a2 chips created from seven different samples (two chips each, one > original sample, the second one with a subpopulation of FACS-sorted > cells). Lacking experience with array analysis I was hoping someone from > the list would be able to identify what the problem might be. > > I've posted the standard QCReport over here (not sure the list can > handle attachments?): > > http://fiamh.info/hsph/QualityControl.pdf > > To a newbie like me it _looks_ fairly okay. From RNA degradation plots I > already knew that E_neg (chip 9) was a problem and it's been discarded; > what seemed odd is the downscaling of all chips (page 3). Also, the > probesets for the cell surface marker used in the FACS sorting exhibited > a very strong foldchange (between 20-30 times, to be expected) whereas > no other probeset showed more than 2-3-fold up- or down-regulation. > > What confuses me is the arrayQualityMetrics report. The variance mean > dependency chart shows quite a bit of variation for the lower ranks: > > http://fiamh.info/hsph/report_sample/QMreport.html > > Mini-script is attached below. I was merely wondering whether the QC is > still within normal (expected) boundaries, or whether they warrant > further investigations. > > > Thanks! > > -- Oliver > > ----8<---- > library(affyQCReport); > library(simpleaffy); > library(arrayQualityMetrics) > > msc.data <- read.affy('covdesc', path='../CEL Files') > QCReport(msc.data, file = "QualityControl.pdf"); > > msc.eset <- call.exprs(msc.data, "mas5"); > arrayQualityMetrics(expressionset=msc.eset, > outdir='report_sample', > force=TRUE, > do.logtransform=FALSE) > > -----8<------ >> sessionInfo() > R version 2.7.1 (2008-06-23) > i386-apple-darwin8.10.1 > > locale: > C/UTF-8/C/C/C/C > > attached base packages: > [1] grid splines tools stats graphics grDevices utils > [8] datasets methods base > > other attached packages: > [1] arrayQualityMetrics_1.6.1 beadarray_1.8.0 > [3] latticeExtra_0.5-1 vsn_3.6.0 > [5] limma_2.14.5 affyQCReport_1.18.0 > [7] geneplotter_1.18.0 annotate_1.18.0 > [9] AnnotationDbi_1.2.2 RSQLite_0.6-9 > [11] DBI_0.2-4 lattice_0.17-8 > [13] RColorBrewer_1.0-2 affyPLM_1.16.0 > [15] xtable_1.5-2 simpleaffy_2.16.0 > [17] gcrma_2.12.1 matchprobes_1.12.0 > [19] genefilter_1.20.0 survival_2.34-1 > [21] affy_1.18.2 preprocessCore_1.2.0 > [23] affyio_1.8.0 Biobase_2.0.1 > > loaded via a namespace (and not attached): > [1] KernSmooth_2.22-22 > > > > >