EDA in Pharmacogenomics With SAS JMP and Windows

EDA in Pharmacogenomics With SAS JMP and Windows WF © OCS Consulting

Contents Introduction Hapmap, JMP and Windows WF Hapmap JMP Windows WF Application Next Steps Conclusions Questions © OCS Consulting

Introduction Dosing Models Model Comparison Modelling EDA Statistical Workflow of a Pharmacogenomic Study Warfarin Subgroup Analysis © OCS Consulting

Clinical Dosing Model Demographic Pharma Clinical From The International Warfarin Pharmacogenetics Consortium. N Engl J Med 2009; 360(8): 753 -764 © OCS Consulting

Pharmacogenetic Dosing Model Demographic Genetic Pharma Clinical From The International Warfarin Pharmacogenetics Consortium. N Engl J Med 2009; 360(8): 753 -764 © OCS Consulting

Model Comparison From The International Warfarin Pharmacogenetics Consortium. N Engl J Med 2009; 360(8): 753 -764 © OCS Consulting

Model Comparison Workflow Impute Split Model Xx 10 x Test Average A Test 10 x Average B Compare Select Model Test 10 x Average Validate © OCS Consulting

Statistical Workflow of a Pharmacogenomic Study 1 From Turner SD, Crawford DC, Ritchie MD. Expert Rev Clin Pharmacol 2009; 2(5): 559 -570 © OCS Consulting

Modelling EDA Modelling / Comparisons: Routine Process • WF compatible EDA : Investigative Process • Smaller (routine) processes Model Extension • Modelling + EDA • WF possibility ? © OCS Consulting

Warfarin From Wadelius M, Pirmohamed M. Pharmacogenomics J 2006; 7(2): 99111. © OCS Consulting

Warfarin 2 From Wadelius M, Pirmohamed M. Pharmacogenomics J 2006; 7(2): 99111. © OCS Consulting

Statistical Workflow of a Pharmacogenomic Study 2 From Turner SD, Crawford DC, Ritchie MD. Expert Rev Clin Pharmacol 2009; 2(5): 559 -570 © OCS Consulting

Subgroup Analysis Presumable more unknown factors are involved that are distributed differently in different populations, e. g. dietary habits. Given this state of knowledge the question remains whether to use one universal dosing algorithm or to use different dosing algorithms for different subpopulations. This subgroup analysis could be extended to other model factors • clinical (diseases / disease states) • pharma (high responder / low responder) Test Test © OCS Consulting

Hapmap, JMP and WWF Hapmap JMP Workflow Windows WF Activities inside a State © OCS Consulting

Hapmap © OCS Consulting

Hapmap 2 © OCS Consulting

JMP Automation JMP Script JMP Workflow Builder © OCS Consulting

JMP 2 © OCS Consulting

Hapmap JMP Workflow © OCS Consulting

Windows WF © OCS Consulting

Windows WF Activities © OCS Consulting

Activities inside a State © OCS Consulting

Start JMP Script in Activity © OCS Consulting

Hapmap JMP WWF Application User Input for State Events User Input for JMP Script JMP Report VKORC 1 Marker rs 9923231 JMP Script Results © OCS Consulting

Hapmap JMP WWF Application 2 © OCS Consulting

User Input for State Events © OCS Consulting

User Input for JMP Script © OCS Consulting

JMP Report © OCS Consulting

VKORC 1 Marker rs 9923231 © OCS Consulting

Results Restriction using Chi-Square Test © OCS Consulting

JMP Script Results © OCS Consulting

Next Steps Different Model factors Different Modelling Techniques Taverna Workflow Warfarin History WWF Custom Activity Tracking Persistence Error Handling Workflow Re-use / Validation © OCS Consulting

Different Model Factors © OCS Consulting

VKORC 1 Marker rs 9923231 Ethnic Clin 1 Clin 2 Clin 3 Pharm 1 Pharm 2 Pharm 3 Clinical Pharma © OCS Consulting

Different Modelling Techniques © OCS Consulting

Detecting and Controlling for Population Stratification From Turner SD, Crawford DC, Ritchie MD. Expert Rev Clin Pharmacol 2009; 2(5): 559 -570 © OCS Consulting

Taverna Workflow Open Source Beanshell Scripts Rshell Scripts (On – off) Data Flow “UI” Command Line App From http: \taverna. sourceforge. net © OCS Consulting

Warfarin History Anti-clotting Agent • high responders • low responders Rodenticide • resistance Rare Familial Diseases • coagulation factor deficiencies © OCS Consulting

Compare Genes of Species or Families © OCS Consulting

WWF Custom Activity © OCS Consulting

Tracking State … 1 2 3 4 © OCS Consulting

Tracking 2 From Chappell D, The Workflow Way. Microsoft. 2009; 1 -27. © OCS Consulting

Persistence From Chappell D, The Workflow Way. Microsoft. 2009; 1 -27. © OCS Consulting

Persistence 2 © OCS Consulting

Error Handling © OCS Consulting

Workflow Re-use / Validation From http: \taverna. sourceforge. net © OCS Consulting

Workflow Validation 2 Validate Stop Workflow © OCS Consulting

Error Handling 2 FTP Unzip Reconnect © OCS Consulting

Hapmap JMP WWF Application 3 © OCS Consulting

Conclusions It is possible to combine a data fetch and JMP scripts in a Windows workflow. The graphical designer is attractive but not a Lego-like activity. With more elaborate modelling or EDA workflows, the Tracking, Persistence and Error handling features probably make it worth considering. © OCS Consulting

Questions ? © OCS Consulting