Cancer is the second leading cause of mortality in the United States and produces enormous public health burden. The overarching goal of cancer genomics is to tailor clinical management according to genetic/epigenetic variations of patients' genomic profile, the so-called personalized medicine. As genome- wide association studies (GWAS), epigenome-wide association studies (EWAS) and expression microarray studies have become standard practices in genomic research, it is also well acknowledged that single genomic platforms may be insufficient to elucidate the biology behind clinical outcomes such as cancer survival. Therefore, considerable interest has emerged in integrating multi-platform genomic data, not only to boost statistical power but also to improve the biological interpretability of these data. Here we propose a novel analytic approach to study cancer survival by integrating multi-platform genomic data. We hypothesize that cancer survival is determined by a biological process from DNA variants to epigenetic methylation and then to gene expression. We will develop a novel algorithm to approach the problem using mediation modeling. The new algorithm is able to delineate the mechanism of cancer survival. We will study glioblastoma multiforme (GBM), a rapidly fatal brain tumor as a disease model. The incomplete understanding in genomics and epigenomics of GBM and its poor prognosis necessitate this innovative genome-wide study of its course. Importantly, the novel analytic approach that will be developed for GBM can also be applied to other cancers. We will pursue the idea of integrative modeling on cancer survival using three concrete study aims: 1) develop methodology to integrate genetics, epigenetics and gene expression data in analyses of cancer survival; 2) perform integrated analyses of DNA variants, gene expression and cancer survival utilizing our newly developed method; 3) perform integrated analyses of DNA variants, DNA methylation and gene expression in cancer survival utilizing our novel method. The ultimate goal is to build an innovative framework capable of integrating multi-platform genomic data to investigate clinical outcome.
Cancer is the second leading cause of mortality in the United States and produces enormous public health burden. Personalized medicine for cancer management requires better understanding of cancer genomics. Given the availability of different genomic data, an integrative view of multiplatform data is critical. This project aims to integrate multiple genomic data sets utilizing a novel discovery algorithm for development of prognostic biomarkers in cancer and may shed light on the biology of this fatal disease.
Showing the most recent 10 out of 17 publications
