Date of Award
5-2024
Degree Type
Dissertation
Degree Name
Ph.D.
Degree Program
Computer Science
Department
Computer Science
Major Professor
Md Tamjidul Hoque
Abstract
This dissertation explores the pivotal role of gene regulatory networks (GRNs) in unraveling the complex regulatory mechanisms that control gene expression, shedding light on cellular operations and the origins of various diseases. The accurate delineation of GRNs is beset with difficulties due to the intricate nature of gene interactions, the prevalence of noise in biological data, and the limitations inherent in current computational models.
This study breaks new ground by employing cutting-edge computational strategies to reconstruct Gene Regulatory Networks (GRNs) from two distinct RNA sequencing methodologies: bulk RNA sequencing and single-cell RNA sequencing. Utilizing ensemble machine learning approaches, the research achieves breakthroughs in deducing GRNs from bulk RNA-seq data, specifically using the DREAM 4 and DREAM 5 datasets. This highlights the effectiveness of combining multiple models to uncover regulatory connections. Furthermore, the research explores the capabilities of graph convolutional networks (GCNs) augmented with a self-attention graph pooling (SAGPool) layer for inferring GRNs. This demonstrates the power of graph-based deep learning in unraveling the intricate, non-linear interactions between genes using the same datasets.
The dissertation expands its investigative scope to single-cell RNA-seq data, offering a novel perspective on the analysis of regulatory networks with unparalleled detail. It presents two distinct types of inferred networks: a comprehensive gene regulatory network, covering a wide array of regulatory interactions, and a transcription factor gene regulatory network, focusing on the critical roles played by transcription factors.
The innovative methodologies developed and deployed here not only advance the domain of GRN inference but also make significant contributions to understanding cellular processes and identifying new therapeutic targets. The efficacious use of ensemble machine learning and graph convolutional networks in GRN inference highlights the potential of these approaches to enrich our understanding of gene regulation. Additionally, applying ensemble encoder-decoder and multiple-layer perceptron techniques to single-cell RNA-seq data marks a significant leap forward in our ability to study cellular heterogeneity and the dynamics of gene regulation. By underscoring the essential nature of GRNs in molecular biology and demonstrating the capability of computational methods to navigate the challenges of their inference, this dissertation lays the groundwork for future breakthroughs in genomics and systems biology.
Recommended Citation
Alawad, Duaa, "Inferring gene regulatory network using Bulk RNA seq and Single-cell RNA seq" (2024). LSU New Orleans Theses and Dissertations. 3415.
https://scholarworks.uno.edu/td/3415
Rights
The University of New Orleans and its agents retain the non-exclusive license to archive and make accessible this dissertation or thesis in whole or in part in all forms of media, now or hereafter known. The author retains all other ownership rights to the copyright of the thesis or dissertation.