Abstract
Genomic profiling produces large amounts of data and a challenge remains in identifying relevant biological processes associated with clinical outcome. Many candidate biomarkers have been identified but few have been successfully validated and make an impact clinically. This review focuses on some of the study design issues encountered in data mining for biomarker identification with illustrations of how study design may influence the final results. This includes issues of clinical endpoint use and selection, power, statistical, biological and clinical significance. We give particular attention to study design for the application of supervised clustering methods for identification of gene networks associated with clinical outcome and provide recommendations for future work to increase the success of identification of clinically relevant biomarkers.
Footnotes
- Received February 13, 2007.
- Accepted February 20, 2007.
- Copyright© 2007 International Institute of Anticaner Research (Dr. John G. Delinassios), All rights reserved