Pilot studies of components of technical variability and practical recommendations for study design. Susan Galloway Hilsenbeck, Baylor College of Medicine, Texas Medical Center Ideally, new microarray experiments are planned with knowledge of the likely magnitude of various sources of experimental variability. Unfortunately, few investigators have conducted the necessary background or pilot studies. Preparatory to undertaking a series of studies of gene expression in models of aging in mice using the Affymetrix GeneChip platform, we have conducted experiments using separately labeled aliquots of the same RNA, using individual and pooled RNA from the same animals, and repeating entire sets of assays on separate days. This has allowed us to explore technical reproducibility and the impact of various "standard" expression estimation algorithms. We have also used components of variance to explore the relative contributions of animal to animal, day to day and residual sources of variability. Finally, we have examined the relative benefits of pooling RNA from separate animals. Pooling has been proposed as a method to reduce total variability, thereby reducing the sample size needed to detect differences between groups. |
Graybill Conference |