Summarization methods and quality problems in Affymetrix microarrays

Abstract

The quality of RNA samples and slide hybridizations are paramount for gene expression studies. Here the focus is on hybridization quality in Affymetrix arrays and a comparison of the results of differential expression analysis for six summarization methods using slides of good and bad quality. The overlap of probes detected as differentially expressed across methods is very small even with good arrays. Slides of inferior quality also significantly change the generated list of differentially expressed genes. Extensive qualitative and quantitative quality control measures should be used prior to downstream analyses of data to ensure early detection and removal of problematic slides. Given the high variability of results across summarization methods it is recommended that different methods are used to conduct analyses and the intersecting results of these are used for further downstream analyses.