|
Abstract:
|
Ovarian cancer is the deadliest gynecologic malignancy with majority patients diagnosed in later stage. Antineoplastic therapeutics is vital to treating serious ovarian patients but have heterogeneous responses. In this study, we present hierarchical mixture model in the Bayesian setting for identifying significant glycoproteins and clusters in response to ovarian cancer treatment. Mass spectrometry (MS) glycolproteome data obtained from Clinical Proteomic Tumor Analysis Consortium, of which 130 ovarian serous carcinoma patients were analyzed. The proposed model have many merits to deal with the special features of MS glycoproteome data, i.e., correcting technical variation, heterogeneity, high percentage missing, strong positive skewness, large proportions of zeros, which make normalize such data using transformations such as powers and logarithms unsuccessful for proper estimate the distribution. Results show that the proposed model identify both different glycoproteins clusters representing subtypes of proteomic profiles of ovarian patients in response to platinum drug treatments, and the up/down glycoproteins representing important biological processes influencing the efficacy of platinum therapeutics.
|
