• Citer

Towards the identification of oligogenic disease variants at exome scale - from oligogenic data to a newprioritisation algorithm.

par Gravel, Barbara
Président du jury Defrance, Matthieu
Promoteur Nowé, Ann;Lenaerts, Tom
Publication Non publié, 2024-09-24

Thèse de doctorat

• Accès en ligne
• Détails
• Contenu
• Statistiques

Résumé :

With the advent of high-throughput sequencing technologies, tremendous progress has been made in understanding the relationship between genotypes and phenotypes. Advances in related computational methods have enabled the development of different prioritization methodsthat help identify in Whole-Exome Sequencing (WES) data which genetic variants are responsible for particular disease phenotypes. However, these methods overlook the fact that a significant proportion of genetic diseases do not follow monogenic inheritance patterns, but are caused by the interaction of variants in a small number of genes. Developping novel computational methods to identifying these more complex combinations of genetic variants, known as oligogenic inheritance, is therefore essential. In this thesis, we build upon existing approaches to detect oligogenic inheritance models to make this analysis possible at the whole-exome level. First, we develop a novel database that collects information on all oligogenic variant combinations reported in the literature. This database not only aggregates existing knowledge but also introduces a standardized framework for assessing the pathogenicity of variant combinations. Using this database, we develop a first oligogenic priorization tool: the High-throughput oligogenic prioritizer (Hop). This predictor integrates pathogenicity scoring, from a machine learning predictor specific to variant combinations, together with disease relevance scoring, based on knowledge propagation in biological networks, to rank variant combinations based on how likely they are to explain a patient’s disease. This tool demonstrates superior performance to existing approaches for ranking oligogenic combinations in exomes. Finally, we investigate the usefulness of these computational tools on real patient data. We apply the Hop predictor on a cohort of patients affected with male infertility, a disease with heterogeneous genetic causes, and investigate the relevance of the prioritized combinations. This analysis validates the ability of Hop to detect oligogenic combinations that were manually identified by clinicians, and also showcases its capacity to identify novel oligogenic signatures in this disease. In summary, this research demonstrates that it is now possible to directly detect disease causing variant combinations in whole exome sequencing data using computational approaches. By introducing a new data repository, computational tools, and analysis protocols, this research opens the way for easier detection and analysis of oligogenic signatures for genetic diseases.