Dr. Tyler Grimes

Abstract:

Gene expression data from RNA-seq experiments provide a measure of biological activity at the cellular level. Differential expression (DE) analyses use these data to identify genes that are up- or down-regulated across groups of interest. Research in systems biology takes this analysis further by gleaning insights into gene regulatory networks through the analysis of gene co-expression networks. Similar to DE analysis, a differential network (DN) analysis is conducted to determine how co-expression patterns change across different groups of interest. A methodology is proposed for predictive modeling in this context that builds on a framework for conducting pathway-based differential network analysis. In an application to cancer gene expression datasets, our goal is to identify driver genes by considering both the functional changes of genes and the clinical relevance of those changes. Functional changes are identified by performing a differential network analysis, which compares the structure of gene-gene associations across different stages of cancer. Clinical relevance of these differential connections is assessed through predictive modeling. Potential driver genes are identified for Neuroblastoma and Breast cancer patient populations, and we identify regulatory pathways that are both differentially connected and whose expression profile is predictive of overall survival.