We're using Affymetrix HG-U133_Plus_2 arrays and our experiment was done so that 8 of our samples were run initially as a test run, followed by running the remaining 74 samples. Since these were two different runs, I'm pre-processing them differently i.e. running rma on them separately. Is there a way to combine the two datasets after their probe level intensities have been normalized within each run since I want to include the initial 8 samples for differential gene expression analysis? If not, what is the best way to include the 8? Please let me know. Thanks! [[alternative HTML version deleted]]

Hi Supriya, if the aim is to analyse the data together, I would try the following: - run rma on all 82 arrays together - produce a quality report using the arrayQualityMetrics package (preferably with the current "devel" version) and check the diagnostics, in particular the heatmap, for how bad the batch effect really is. - if there is no strong confounding between your two batches and the biological factor of interest, find differentially expressed genes using limma, keeping the batch as a covariate to absorb the (average) batch effect. - for the hits, go back to the raw data (e.g. via scatterplot or heatmap) to double-check that there are no batch-related artifacts. This seems to be an instance of where lazy experimental design has to be compensated by a more difficult data analysis, and you might want to have a word with whoever designed this experiment about that. best wishes Wolfgang On Sep/10/10 10:53 PM, Supriya Munshaw wrote: > We're using Affymetrix HG-U133_Plus_2 arrays and our experiment was done so that 8 of our samples were run initially as a test run, followed by running the remaining 74 samples. Since these were two different runs, I'm pre-processing them differently i.e. running rma on them separately. Is there a way to combine the two datasets after their probe level intensities have been normalized within each run since I want to include the initial 8 samples for differential gene expression analysis? If not, what is the best way to include the 8? > Please let me know. > Thanks! > > [[alternative HTML version deleted]] > > _______________________________________________ > 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 -- Wolfgang Huber EMBL http://www.embl.de/research/units/genome_biology/huber

As an alternative to limma in case of batch effect, you may try pamr.batchadjust(), works quite well for me, see also eg: http://www.biomedcentral.com/1755-8794/1/42 Cheers, Michal -----Mensaje original----- De: bioconductor-bounces at stat.math.ethz.ch en nombre de Supriya Munshaw Enviado el: vie 10/09/2010 10:53 Para: bioconductor at stat.math.ethz.ch Asunto: [BioC] Combining data from runs We're using Affymetrix HG-U133_Plus_2 arrays and our experiment was done so that 8 of our samples were run initially as a test run, followed by running the remaining 74 samples. Since these were two different runs, I'm pre-processing them differently i.e. running rma on them separately. Is there a way to combine the two datasets after their probe level intensities have been normalized within each run since I want to include the initial 8 samples for differential gene expression analysis? If not, what is the best way to include the 8? Please let me know. Thanks! [[alternative HTML version deleted]] _______________________________________________ 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