“Bayesian Kernel Based Models and Analysis of High-dimensional Multiplatform Genomics Data”
Abstract: In recent years we have seen rapid development of new technologies for genome-wide assays. With increasing reliability and affordability of microarray and next-generation sequencing, patient care decisions are now customized based on the diverse genetic and epigenetic alterations of a disease for a specific individual. Moreover, nowadays the scale of omcis studies has expanded to measure and include multiple genomic features on a single patient, like gene expression, DNA methylation, gene mutation, copy number variation, promoter binding and protein expression. Combining and modeling multiple genome features coming from different data platforms is a big conceptual challenge and practical hurdle.
In this talk we propose to develop statistical nonlinear models to integrate genomic data from multiple platforms. Our models can incorporate the fundamental biological relationships that exist among the data obtained from different platforms and produce more accurate understanding of the functional responses. The proposed models are developed on the basis of Bayesian trees and Bayesian kernel machine models. Our methodologies are highly flexible in exploring, extracting, and analyzing complex biological systems and datasets from heterogeneous platforms. Combining all available genetic, pathological, and demographic information can dramatically improve the nature of clinical diagnosis and treatment of several human diseases.
For inquiries contact Chengjie Xiong.