MAQC II Using JMP Genomics for Predictive Modeling
JMP Genomics development group leader Russ Wolfinger and Padraic Neville (SAS Analytics R&D) recently attended the 8th Face-to-Face meeting of the MicroArray Quality Control (MAQC) Consortium at the FDA in Rockville, MD. Results of the first phase of MAQC were published in the journal Nature Biotechnology in late 2006 and focused on validation of microarray technology platforms.
Predictive modeling is the focus of MAQC’s second phase (commonly called MAQC II), and much of the current work has involved creating and testing predictive modeling for several different data sets. This project has prompted lots of updates to JMP Genomics predictive modeling processes by Russ, Padraic, Pei-Yi Tan (also from the SAS Analytics group), and other developers from the JMP Genomics group. New for release 3.2 are Learning Curves and various enhancements to Cross-Validation Model Comparison, including drill-downs to display more detailed cross-validation results.
The genomics connection between scientists at the FDA’s National Center for Toxicological Research (NCTR) and Russ’ development teams at SAS dates back several years. A project to integrate JMP Genomics with the ArrayTrack database developed by Dr. Weida Tong’s group at NCTR has been completed under a Cooperative Research and Development Agreement with NCTR. As a result of the integration work, scientists using both programs can easily move data back and forth between ArrayTrack and JMP Genomics, taking advantage of the analytic and data storage capabilities of both platforms. Dr. Weida Tong is the Director of the FDA/NCTR Center for Toxicoinformatics, and his group, including Dr. Leming Shi and others, coordinates many aspects of the MAQC project.