1. Microarray Data Analysis Data quality assessment and normalization for affymetrix chips

2. 2 Outline Visualization and diagnostic plots Normalization

3. 3 Microarray studies life cycle Here we are

4. Looking at microarray data Diagnostic Plots Was the experiment a success?

5. 5 Image plots for affymetrix chips

6. 6 MA-plot for GeneChip arrays (1 color) MT WT RMA MT intensity for each probe set RMA WT intensity for each probe set aRNA M Log 2 (MT) - Log 2 (WT ) A Log 2 (MT*WT) / 2 (signal strength)

7. 7 Box plots

8. 8 Density plots

9. 9 Digestion plots

10. Preprocessing affy chips

11. 11 Preprocessing steps Computing expression values for each probe set requires 3-steps  Background correction  Normalization  Probe set summaries

12. 12 Most popular approaches Affymetrix’s own MAS 5 or GCOS 1.0 algorithms RMA (Robust Multichip Analysis)  Irizarry, Bolstad, Collin, Cope, Hobbs, Speed (2003), Summaries of Affymetrix GeneChip probe level data. NAR 31(4):e15 dChip http://www.dchip.org: Li and Wong (2001). Model-based analysis of oligonucleotide arrays: expression index computation and outlier detection. PNAS 98, 31-http://www.dchip.org

13. 13 MAS Background correction  E j = PM j - MM j * where MM j * is chosen so that E j is non-negative Normalization  Scale so that mean E j is same for each chip Probe Set Summary  log(Signal Intensity) = TukeyBiweight(log Ej)

14. 14 RMA- Background correction Ignore MM, fit model to PM  PM = Background (N(0,  2 ) + Signal (Exp(  ))

15. 15 RMA-Normalization Force the empirical distribution of probe intensities to be the same for every chip in an experiment The common distribution is obtained by averaging each quantile across chips: Quantile Normalization

16. 16 One distribution for all arrays: the black curve

17. 17 RMA: Probe set summary Robustly fit a two-way model yielding an estimate of log2(signal) for each probe set Fit may be by  median polish (quick) or by  Mestimation (slower but yields standard errors and good quality RMA reduces variability without loosing the ability to detect differential expression

18. 18 MA plots before and after RMA

19. 19 Summary Microarray experiments have many “hot spots” where errors or systematic biases can apper Visual and numerical quality control should be performed Usually intensities will require normalisation  At least global or intensity dependent normalisation should be performed  More sophisticated procedures rely on stronger assumptions  Must look for a balance