Dear Peter did you already generate a quality report with 'arrayQualityMetrics'? Also, I assume you (or the biologist who designed the experiment) knows about some controls, for which it is already known what they should be doing; or is able to test some of the results using an independent assay. The dimness of your arrays however suggests that there could be a problem with data quality that is not easy to fix by data analysis. Best wishes Wolfgang Il giorno Dec 7, 2012, alle ore 10:44 AM, Peter Davidsen <pkdavidsen at="" gmail.com=""> ha scritto: > Dear List, > > I'm analysing some one-color microarray data generated using a custom > designed Agilent array (their 8 x 60K platform). > When I compare control samples to treated ones most (i.e. >80%) of the > differentially expressed transcripts are up-regulated. This pronounced > up-regulation is independent of type of treatment. > I have never before experienced such a quantitative difference in the > number of up- and down-regulated transcripts. Furthermore, I have > tried to analyse relevant datasets in the GEO that mimics my study > design in terms of treatment regime, duration of treatment ect. These > analyses--all in closely related species--suggest that the fraction of > up and down-regulated transcripts should be roughly the same. > The QC reports generated by the Agilent Feature Extraction Software > indicate that the data quality should be fine. Also a few basic > boxplot before and after normalization haven't raised my suspicion. I > do, however, find that the median signal intensity for each sample is > significantly lower than what I've seen in the past with the same > platform (although targeted against another related species). > I have tried to normalize my data using both quantile and vsn, > respectively, with similar result. I have also tried to filter my > dataset using different intensity filters - again with similar result. > And finally, I also tried using both limma and SAMR for the > statistics. > > Have anyone by any chance experienced something similar, and how did > you deal with this issue of many siggenes going in one direction? > > Many thanks, > Peter > > _______________________________________________ > 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

Dear Peter did you already generate a quality report with 'arrayQualityMetrics'? Also, I assume you (or the biologist who designed the experiment) knows about some controls, for which it is already known what they should be doing; or is able to test some of the results using an independent assay. The dimness of your arrays however suggests that there could be a problem with data quality that is not easy to fix by data analysis. Best wishes Wolfgang Il giorno Dec 7, 2012, alle ore 10:44 AM, Peter Davidsen <pkdavidsen at="" gmail.com=""> ha scritto: > Dear List, > > I'm analysing some one-color microarray data generated using a custom > designed Agilent array (their 8 x 60K platform). > When I compare control samples to treated ones most (i.e. >80%) of the > differentially expressed transcripts are up-regulated. This pronounced > up-regulation is independent of type of treatment. > I have never before experienced such a quantitative difference in the > number of up- and down-regulated transcripts. Furthermore, I have > tried to analyse relevant datasets in the GEO that mimics my study > design in terms of treatment regime, duration of treatment ect. These > analyses--all in closely related species--suggest that the fraction of > up and down-regulated transcripts should be roughly the same. > The QC reports generated by the Agilent Feature Extraction Software > indicate that the data quality should be fine. Also a few basic > boxplot before and after normalization haven't raised my suspicion. I > do, however, find that the median signal intensity for each sample is > significantly lower than what I've seen in the past with the same > platform (although targeted against another related species). > I have tried to normalize my data using both quantile and vsn, > respectively, with similar result. I have also tried to filter my > dataset using different intensity filters - again with similar result. > And finally, I also tried using both limma and SAMR for the > statistics. > > Have anyone by any chance experienced something similar, and how did > you deal with this issue of many siggenes going in one direction? > > Many thanks, > Peter > > _______________________________________________ > 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