1. 1 Application of DOE Methodology to SNP Assay Development Geetha Rajavelu Gretchen Kiser 2002 Quality & Productivity Research Conference Tempe, AZ June 5-7, 2002

2. 2 Presentation Outline Background Info Overview of SNP Assay Identifying Variables for Screening Experiment Screening Experiment Results Optimization Work Overall Performance Improvements Concluding Remarks

3. 3 Background Info In Q3 2000, CodeLink slides were affected by unexplained background. All aspects of slide manufacturing and assay processes were examined for root causes. An effort was launched to thoroughly characterize all processes (with no apriori assumptions).

4. 4 SNP Assay Modules Target Amplification Target Pooling & Fragmentation SBE Thermal Cycling SBE Slide Washing, SA-Alexa Labeling & Post-labeling Slide Washing Drying Scanning

5. 5 S = major allele s = minor allele PCR clean-up & Fragmentation Add reaction mixture to Bioarray Allele-Specific Extension Washing & Secondary Labeling SNPSNP PCR DNA genomic primers SNPSNP SNPSNP Whole target DNA SNPSNP Fragmented target S S SNPSNP SsS Target hybridizes to probes on Bioarray SsS Extension reaction on Bioarray Mediated by a DNA polymerase and labeled nucleotides Allele on Bioarray ready to scan SNP Assay Overview

6. 6 Screening Experiment Info Objective: –To evaluate the contribution of any one variable, and the interactive contribution of any two variables of the SBE Thermal Cycling module, to the assay performance, as measured by the Signal/Noise ratio, the Call Rate and Accuracy, the Mean Pad IOD, and the Mean Blank Pad IOD. Preliminary Info: –Factor ranges based on single-variable experiments –Nuisance Factors: Dispense Slide Batches, Thermal Cyclers Design Details: –2 12-6 fractional factorial design –Involved 65 runs, 2 slides/run –Took around 9 days to complete experiment

7. 7 Identifying Variables for the Allele-Specific Extension Reaction Thermal Cycler Number of Cycles (A) Extension Time (B) Denaturation Time (C) Extension Temperature (D) Denaturation Temperature (E) Reaction Mixture Target Concentration (F) Enzyme Concentration (G) tNTP Concentration (H) Salt1 Concentration (J) Salt2 Concentration (K) Buffer Concentration (L) Buffer pH (M)

8. 8 Selecting Significant Factor Effects

9. 9 Screening Experiment Analysis Results

10. 10 Interaction Effects

11. 11 Interaction Effects

12. 12 Interaction Effects

13. 13 Parameters shaded in green represent subset chosen for inclusion in the optimization experiment. Screening out Variables for Optimization Phase

14. 14 Response Surface Experiment Info Objective: –To determine factor settings which optimize the assay performance, as measured by the Signal/Noise ratio, the Call Rate and Accuracy, the Mean Pad IOD, and the Mean Blank Pad IOD. Variables: -- # of Cycles (A)-- Enzyme Conc (E) -- Extension Time (B)-- tNTP Conc (F) -- Extension Temp (C)-- Salt2 Conc (G) -- Denaturation Temp (D) Design Details: –Screening design augmented with axial runs using face-centered cube & D-Optimality criterion (24 runs total)

15. 15 Response Surface Modeling Results

16. 16 Numerical Optimization Criteria

17. 17 Numerical Optimization Results

18. 18 Optimal Region

19. 19 Optimal Region

20. 20 Optimal Region

21. 21 SNP Assay Modules – Characterized & Optimized Target Amplification Target Pooling & Fragmentation (2 3 factorial; face-centered design in 3 factors) SBE Thermal Cycling (2 12-6 fractional factorial; face-centered design in 7 factors) SBE Slide Washing, SA-Alexa Labeling & Post-labeling Slide Washing (2 7-2 fractional factorial; face-centered design in 5 factors) Drying Scanning

22. 22 Overall Performance Improvements Group 1 DOE specified master mix DOE specified thermocycle program Low salt TNT mix Low SA-Alexa 532 dilution Optimized Staining/Washing conditions Group 2 DOE specified master mix DOE specified thermocycle program High salt TNT mix High SA-Alexa 532 dilution Optimized Staining/Washing conditions Current Protocol Group 3 Current master mix Current thermocycle program Current TNT mix Current SA-Alexa 532 dilution Current Staining/Washing conditions

23. 23 Higher Call Rate with DOE-derived Conditions

24. 24 Higher Call Accuracy with DOE-derived Conditions

25. 25 3X Improvement in Probe Intensity

26. 26 3X Improvement in Signal/Noise

27. 27 Lower Background with DOE-derived Conditions

28. 28 Group 1 Group 2Current Protocol Group 3

29. 29 Concluding Remarks All experiments were planned and executed using Project Management approach (risks/assumptions, scheduling, etc.) Sequential approach to experimentation is critical to studying complex processes. Able to beat all odds!! The probability of successfully completing an experiment is inversely proportional to the number of runs, i.e.,1/65.

30. 30 Acknowledgements SNP Assay Development Team Project Management Team