ABSTRACT
Understanding genomic structural variation such as inversions and translocations is a key challenge in evolutionary genetics. In this paper, we tackle this challenge by developing a novel statistical approach to comparative genetic mapping. The procedure couples a Hidden Markov Model with a Genetic Algorithm to detect large-scale structural variation using low-level sequencing data from multiple genetic mapping populations. We demonstrate the method using five distinct crosses within the flowering plant genus Mimulus. The synthesis of data from these experiments is first used to correct numerous errors (misplaced sequences) in the M. guttatus reference genome. Second, we confirm and/or detect eight large inversions polymorphic within the M. guttatus species complex. Finally, we show how this method can be applied in genomic scans to improve the accuracy and resolution of Quantitative Trait Locus (QTL) mapping.
AUTHOR SUMMARY Genome sequences have proved to be a critical experimental resource for genetic research in many species. However, in some species there is considerable variation in genomic organization, making a single reference genome sequence inadequate. This variation can cause issues in interpreting genomic signals, such as those coming from trait mapping. We introduce a new statistical method and computational tools that use linkage information to reorganize a single reference genome to 1) repair genome assembly errors, and 2) identify variation between individuals or populations of the same species. Using this method we can create a new genome order that improves upon the reference genome. We apply this method to five crosses among plants in the Mimulus guttatus species complex. In this system we detect eight large chromosomal inversions and improve the resolution of a trait mapping study. This work highlights the utility of our method, and indicates how others studying diverse species might use them to improve their own research.
