Year: 2026
Type: dissertation
Abstract: The rapid advancement of high-throughput technologies in the genomics era has driven the generation and availability of vast volumes of biological data, ranging from gene expression studies to summary statistics from large-scale Genome-Wide Association Studies (GWAS).The effective utilization of this data to derive valid and generalizable conclusions requires efficient computational methodologies characterized by both accuracy and speed.Despite significant progress in recent years, a gap between methodological development and the practical tools available to the research community remains, particularly regarding the meta-analysis of gene expression data and genetic association studies, as well as the utilization of summary statistics.The present PhD thesis falls within the broader framework of Bioinformatics and Computational Biology, focusing on the development and implementation of computational methods for the analysis and meta-analysis of gene expression data and GWAS.The thesis was motivated by the observation that, while numerous software tools exist, there remains a lack of comprehensive methodological frameworks that allow researchers to flexibly apply both traditional and modern approaches in a computationally efficient manner, while adapting them to the specific characteristics of each dataset.Primarily, this thesis aims to bridge this gap through the development of new algorithms and software for the meta-analysis of both gene expression data and genome wide association studies, with particular emphasis on leveraging summary statistics.Significant weight is placed on integrating methods that, despite their theoretical value, have not yet found widespread application in existing tools, such as Bayesian meta-analysis approaches.It also focuses on supporting multiple analysis scenarios that reflect the heterogeneity and complexity of the data.In parallel, the thesis addresses the issue of summary statistics imputation, explicitly accounting for linkage disequilibrium information from multiple reference panels, aiming to improve coverage and the reliability of conclusions.Beyond establishing a comprehensive methodological framework, an equally important objective is the algorithmic optimization of the proposed approaches.This need arises from the fact that modern data, whether from large consortia with hundreds of thousands of participants in genomewide association studies or extensive gene expression datasets, renders traditional method implementations limited in their application.Therefore, the thesis also focuses on accelerating metaanalysis and imputation processes so they can be executed within reasonable timeframes in modern computational environments, as well as on increasing the precision of statistical estimates under realistic conditions of heterogeneity and in multivariate analysis scenarios.
Author: Μανιός, Γεώργιος Α.
Institution: University of Thessaly
Language: Greek
Subfield: Genetics
Domain: Life Sciences
Open Access status: green
Award: G7618027526
OpenAlex ID: W7223480622
