Risk diplotypes promoting recombination error in hybrid and non-hybrid human lineages

Abstract

PRDM9 is a key player in meiotic recombination, but the consequences of variation in its binding landscape remain poorly understood. This thesis investigates how single nucleotide variation predisposes the human genome to structural rearrangement via non-allelic homologous recombination. Through population genomic analyses, we first demonstrate that SNPs creating or disrupting PRDM9 binding site motifs are widespread. We show that these SNPs, particularly those that disrupt PRDM9 binding sites, are subject to purifying selection and are concentrated at the edges of CNVs. By expanding this logic to ancient hybridisation events, we demonstrate that archaic segments in the modern human genome are hotspots of genomic instability, likely due to the substantial divergence in PRDM9 landscapes between modern humans and archaic hominins. These archaic segments possess distinct genomic signatures, including an enrichment of duplications within the segments along with deletions flanking them, consistent with our proposed model. Furthermore, the depletion of archaic introgression in regions of high recombination suggests that this genomic instability may have contributed to the gradual erosion of archaic sequences from our genome. Thus, the archaic legacy in modern humans is a dynamic and potentially disruptive force, which actively shapes our genome through ongoing structural mutations.