Utilisation du profilage statistique pour déchiffrer les contacts cachés de la chromatine résultant de séquences répétées

Genomes are not static entities; they evolve and acquire various repeated elements such as retrotransposons, duplicated genes and tandem repeats. These elements, while crucial to genome function and evolution, pose a significant challenge to modern genomics due to the limitations of next-generation sequencing (NGS) technologies. NGS produces short reads that cannot be unambiguously mapped to repetitive regions, resulting in their exclusion from most analyses. This omission leaves a critical gap in our understanding of genome structure and function, as well as complex biological processes involving repetitive elements. This thesis tackles this challenge head-on by developing Hicberg, an algorithm that relies on statistical inference and pseudo-random generators to predict the genomic locations of reads from repetitive elements. Hicberg is compatible with various types of paired-end sequencing data, including Hi-C, MNase-seq and ChIP-seq, making it a versatile tool for genomics research. The effectiveness of the method is demonstrated throughout this thesis, through rigorous testing on a reference dataset, and presents its application in the model organism Saccharomyces cerevisiae. In yeast, the reconstruction of Hi-C and ChIP-seq data using Hicberg reveals the impact of retrotransposons in the positioning of cohesin, a protein complex essential for the formation of chromatin loops. In addition, Hicberg identifies new interactions between retrotransposons and the enigmatic 2-micron plasmid in yeast.Furthermore, under stress conditions, Hicberg discovered specific interactions between ribosomal DNA and other genomic loci, suggesting a potential role in the regulation of stress response pathways. These results underline Hicberg's ability to uncover previously hidden molecular relationships, and highlight the possibility of applying this approach to larger, more complex genomes with even higher proportions of repeated elements. By reconstructing genomic signals from repeated regions, Hicberg offers a more complete and precise view of genome organization and plasticity.