Bayesian Functional Models for Biomarker Discovery in Clinical Proteomics Study

• Activity Number: 603
• Type: Topic Contributed
• Date/Time: Thursday, August 2, 2012 : 8:30 AM to 10:20 AM
• Sponsor: Section on Government Statistics
• Author(s): Xia Wang*+
• Companies: University of Cincinnati
• Keywords: Bayesian Functional regression ; Biomarker ; Proteomics

Abstract

The study aims at significantly improving biomarker detections in proteomics analysis by modeling the protein profile as a whole using functional regression approaches. Distinct functional surfaces are used in analyzing and modeling various sources of expression variations. A Bayesian mixture prior model is used to cluster the proteins based on the difference between control and case samples. Developing this model both significantly increases the biomarker detection power in practice and contributes to the methodology development of Bayesian functional modeling. The proposed model is applied to the experimental data from the Clinical Proteomic Technology Assessment for Cancer (CPTAC) team and the E.coli dataset in Finney et al. (2008). The proposed project has very broad applications to other scientific fields that are facing similar massive data and need strong feature detection algorithms, like genomics, fMRI and environmental sciences.