Factors influencing genetic variation in wild mice

The mouse is a premier model organism for mammalian biological research. They have been instrumental in studies illuminating processes of fundemental importance to genetics including the control of recombination and the process of speciation. However, most of these studies are based on laboratory mice, which are an artificial population with a complex relationship to wild mice. Studies of mammalian biology would benefit from understanding how history has shaped the genome of the present day mouse. In this thesis, I explore factors that contribute to patterns of genetic variation in wild mice, paying particular attention to the process of recombination. I use whole genome sequencing data from three wild populations: 20 French M. m. domesticus; 20 Taiwanese M. m. castaneus; and 10 Indian M. m. castaneus. In addition, I use 13 classical laboratory strains and 6 wild derived inbred strains. These data show that the French and Taiwanese populations have been through recent, severe population bottlenecks, are recipients of considerable migration, and are partially inbred, with about 15-20% of the genome recently homozygous. All of these features are absent from the Indian mice. Signatures of selection reveal that the Prl gene in Indian mice was the result of a full sweep originating from a highly diverged species of mouse. In terms of recombination, the linkage-disequilibrium based rate map of the French and Taiwanese mice is shown to be highly punctuate, while the Indian rate map is comparatively flat, and devoid of population level hotspots. The recombination landscape is shown to dominate nucleotide substitution, notably in the Indian mice, where it reverses a 2:1 strong (C/G) to weak (A/T) mutation bias in favour of a 2:1 weak to strong fixation rate. Dozens of long-extinct, and active, PRDM9 motifs are identified through a comparative genomic approach, which also reveals a highly localized picture of the distribution of GC-biased gene conversion. Separately, a low-coverage reference-panel free read-aware genotype imputation method named STITCH is presented, and shown to accurately impute genotypes in outbred laboratory mice and humans.