INTEGRATIVE APPROACHES FOR THE STUDY OF COMPLEX HUMAN DISEASE

Abstract

The opportunities that lead to the detection, treatment and prevention of diseases oftentimes require a systemic understanding of what cellular changes accompany the disease. High- throughput approaches such as the microarray and RNA-sequencing have empowered researchers to study the behavior of thousands of genes within a cell. The integration of this data across pathological states and multiple experiments presents many opportunities to improve our understanding of human diseases. This thesis represents the work of two projects focused on integrating high throughput data to identify genes associated with a disease. The first project seeks to understand the changes in expression that occur during oncogenesis. By integrating gene expression data across three histological mammary tissue states (normal, adenoma, and carcinoma) we have identified three distinct patterns of gene expression that emerge during the progression of a tumor. We show that these disease-progression associated genes represent known cancer-related pathways. The second project utilizes Naïve Bayesian machine learning to predict novel immune functional relationships by distilling the data from a large compendium of high-throughput gene expression data. We built an interactive web resource for exploring the relationships within these generated networks. Furthermore, we utilized these networks to successfully predict genes associated with several immune diseases.