| Recommendations for experimental design of standard experiments (2008) | |||||||||||||||
Abstract | |||||||||||||||
| • Define the goal of your experiment in a quantitative way: Make an appropriate structural compromise between the most informative and most robust set-up for your experiment. • For cDNA microarrays comparisons made on the same slide are always more accurate than comparisons linked via several slides via common references. For experiments with only few comparisons of interest it is best to choose a design, where these comparisons are performed directly. • For experiments with many comparisons of interest a design including all possible comparisons often becomes unfeasible. There is usually a choice between several different designs, which should be carefully considered. • The most common design for two-color (competitively hybridized spotted) arrays is the ‘reference design’: each experimental sample is hybridized against a common reference sample. Although this effectively means that only one sample of interest is hybridized per chip, the reference design has several practical advantages over more efficient designs: extends easily to other experiments, if the common reference is preserved, facilitates inter-comparison of cross-array data, is robust to multiple chip failures, reduces incidence of laboratory | |||||||||||||||
Publication details | |||||||||||||||
| |||||||||||||||