Résumé : This thesis addresses a critical need in understanding genetic diseases influenced by complex combinations of rare variants, dispersed across multiple genes. Traditional genetic studies often focus on single-gene mutations, which leaves much of the genetic landscape unexplored, particularly for diseases where multiple genetic or environmental factors play a role. This research focuses on aggregation tests, a powerful approach to detecting associations between groups of variants and traits by combining multiple genetic signals into a single, interpretable score. After extensive benchmarking of 59 existing methods, this work introduces Excalibur, a novel aggregation test that demonstrates robust performance across a variety of genetic models. Additionally, this thesis presents DoAggregate, a pipeline designed to facilitate rare-variant analysis across cohorts at the level of genes and pathways. The application of DoAggregate to a dataset of primary lymphedema cases offers new insights into oligogenic inheritance and genetic modifiers, marking a step forward in our understanding of complex genetic conditions.