Experimental Design

1. Reaching a balance between statistical power and available finances Experimental Design

2. Cost of Microarrays • Glass arrays €250 - €400 • Affymetrix arrays €700 - €900

3. Experimental Design • Choice of microarray • Hybridization design • Number of replicates • Dye-bias • RNA samples

4. Choice of Microarray • Glass slide v Affymetrix • In house v Commercial • Oligonucleotide v cDNA • Focal v Global

5. Two- Colour Microarray Designs • Reference design • Loop design

6. Classical Reference Design

7. Common Reference Design

8. Loop Design

9. Number of Replicates • Technical replicates • Biological replicates • Increasing the number of replicates = • Increases statistical power. • Increases cost of the microarray experiment. • Increases animal costs.

10. Dye Bias • Cy3 and Cy5 can bias binding to particular array spots. • Include a dye-swap of each array to identify and remove these problems. • Doubles the number of microarrays required. Cy3 Cy5

11. RNA Samples • Tissue/cell type • Time course • Quality of RNA • Quantity of RNA

12. good quality RNA poor quality RNA Quality of RNA 1) Extract RNA using Trizol 2) Purify RNA using Qiagen RNeasy column 3) QC RNA using Aligent BioAnalyzer

13. AAAA 3’ 5’ Total RNA First strand cDNA synthesis 5’ cDNA AAAA 3’ TTTT-T7 5’ 3’ Second strand cDNA synthesis Transcription template AAAA-T7 5’ TTTT-T7 3’ In vitro transcription (incorporation of amino allyl UTP) UUUU 5’ Antisense RNA UUUU 5’ UUUU 5’ UUUU 5’ Labelling Hybridization Linear Amplification of RNA

14. Conclusions (2) • Need to balance statistical power and cost. • Need to reduce variation and increase the statistical power of the experiment by: • Design of the experiment • Replicate spots • Technical replicates • Biological replicates • Dye-swap • Good quality RNA is essential.