U.S. Department of Energy

Pacific Northwest National Laboratory

Experts Inundated with Data: The Biomarker Problem

Principle Investigator: 

A primary challenge facing biomarker scientists is the need to integrate expert-driven and data-driven approaches to biomarker discovery. To date, exclusive reliance on either approach has proven to be unsatisfactory. Researchers will tackle this problem using an extensive set of proteomic and transcriptomic data already available from a mouse model of chronic obstructive pulmonary disease.

The initial approach will explore the use of expert-driven approaches to weight the -omic data prior to statistical, data-driven analyses such as hierarchical clustering, network analysis, or support vector machine analysis. Once a suitable process has been defined, researchers will test the domain independence of that process or algorithm by using it to analyze a distinct dataset, to be determined at that time.

The overall objective of this research is to develop a powerful tool for successfully applying expert knowledge to data-driven analysis of large complex datasets, so as to extract the maximum knowledge and insight. The abstract principles developed can be applied to any type of complex, high density data, including problems in environmental remediation, national defense, and energy production.

Project Staff: 
Jason McDermott, Bill Cannon, Bret Cannon, Ryan Hafen, Melissa Matzke, Ann Miracle, Sam Payne, Karin Rodland, Jing Wang, Bobbie-Jo Webb-Robertson
PDF icon SDI_Inundated_experts_Flyer.pdf1.55 MB
| Pacific Northwest National Laboratory