All ETDs from UAB

Advisory Committee Chair

Jake Y Chen

Advisory Committee Members

Nengjun Yi

Louis S Bridges

Zechen Chong

Robert S Welner

Jake Y Chen

Document Type

Dissertation

Date of Award

2020

Degree Name by School

Doctor of Philosophy (PhD) Heersink School of Medicine

Abstract

With rapidly increasing novel discoveries of associations between genomic data and complex disease phenotypes, the translation of those associations into knowledge and the identification of critical molecular mechanisms are essential to guild clinical decisions such as the diagnosis of symptomatic individuals, the prediction of disease risk, reproductive genetic counseling, and determining pharmacogenetic profiles for treatment. To decipher the genome of complex diseases, gene-centric analyses focus on detecting molecular mechanisms lead by driver genetics variants, abnormally expressed genes and high-abundance proteins through gene module construction. Additionally, the gene module construction approaches implement systems biology analysis to reveal the causative co-expressed gene associations. Even though the traditional gene co-expression modules make huge progress in identifying module eigengenes in neuroscientific applications, several arguments are raised regarding homogeneous network in use, lack of biological relevance, and non-sharable gene members among modules which impede biological interpretation and scalability in real-world applications. The thesis presents a new framework in network-based gene module construction and visualization with three advanced aspects. (i) integration of experimental data (transcriptome) and knowledge-based data (interactome and gene ontologies) in network construction, (ii) layout optimization considering multiple biological factors, and (iii) Exploration of gene signals and insights in the constructed Terrain Knowledge Mappings (TKMs). This thesis is based on a collection of published works in collaboration with several teams, and I will divide our contributions into three major parts. In Part 1 “Data curation”, we focus on data cleaning and database integration to provide a pre-knowledge basis in TKMM construction. Specifically, we will introduce three databases, (i) pathways, annotated-lists and gene signatures electronic repository (PAGER 2.0), (ii) drug-protein connectivity map (DMAP), and (iii) biomedical entity extension ranking and exploration (BEERE). In Part 2 “Network-based module construction”, we address two novel algorithms applied in network-based gene module construction. One algorithm called “Weighted in-Path Edge Ranking for Biomolecular Association Networks (WIPER)” enables Protein-Protein Interactions prioritization by evaluating initially edge weights and global topological structure. The other algorithm “distance-bounded energy-field minimization algorithm (DEMA)” is for quantitatively optimizing network layout with biological properties such as gene weights, gene-gene relationships and gene associations. In Part 3 “Visual analytics”, we present a systems biology analysis pipeline in constructing TKMs and discovering disease markers using visualization technique "GeneTerrain". The TKMs in a Graft Versus Host Disease study reveal specific gene markers validated in clinical tests. Finally, I summarize our contributions and demonstrate further directions in assisting and accelerating biomedical discovery cycles in computerized translation of biological data into informative insights in complex diseases.

Share

COinS