EcoSta 2021: Start Registration
View Submission - EcoSta2021
A0749
Title: Nonparametric Bayesian functional clustering for breast cancer disparities Authors:  Wenyu Gao - University of North Carolina at Charlotte (United States) [presenting]
Inyoung Kim - Virginia Tech (United States)
Wonil Nam - Virginia Tech (United States)
Wei Zhou - Virginia Tech (United States)
Abstract: It has been found that different incidence and mortality rates for breast cancer exist among various racial populations. For instance, Caucasian women are more likely to develop breast cancer than African American women. To study these disparities, surface-enhanced Raman spectroscopy (SERS) has been conducted to provide biomolecular fingerprint information. Extracellular SERS signals from each cell type were measured by a practical high-performance SERS device. However, large intraclass variations exist due to cellular and additional cancerous heterogeneity. Therefore, we need to reduce the amount of noise information and make each group distinguishable. The noises exist in two directions: the large number of heterogeneously behaved signal curves, as well as the massive change points on each curve. To study the differences between two types of triple-negative breast cancer cell lines at the molecular level, we performed functional cluster analyses and change point selection methods on the massive nonlinear curves of signals versus Raman shifts. Thus, we propose a nonparametric Bayesian functional clustering and change point selection method via weighted Dirichlet process mixture (WDPM) modeling, together with conditional Laplace prior. The proposed method is named WDPM-VS for short, and it will greatly outperform its comparison methods. Based on this proposed method, we identified important wavelengths that will explain the racial disparities.