Integrative Analysis of Multi-modal Correlated Imaging-Genomics Data in Glioblastoma

We propose a method to integrate high-dimensional genomics datasets across multiple platforms with multiple imaging outcomes. This new statistical framework uses a hierarchical model to integrate biological relationships across platforms to identify genes that associate with multiple correlated imaging outcomes. Our two-stage hierarchical model uses the information shared across the platforms and thus increasing the predictive power to identify the relevant genes. We assess the performance of our proposed method through simulation and apply to data obtained from the Cancer Genome Atlas Glioblastoma Multiforme dataset. Our proposed method discovers multiple copy number and microRNA regulated genes that are related to patients' imaging outcomes in glioblastoma.