Genome Mining: Evaluating Germplasm Core Collections for Population Structure and Naturally Occurring Admixture

Following the completion of the rice genome reference assembly and structural annotation, fine mapping of QTL, positional cloning, mutant studies, and comparative genomics have elucidated gene function and biochemical pathways contributing to many key phenotypes. To improve essential agronomic traits such as yield, grain quality, abiotic stress tolerance, and disease resistance, candidate loci can be sequenced across a large panel of genotypes to discover naturally occurring alleles that are already available in so-called core collections as donor lines for breeding improved varieties. However, rice is characterized by a deep population structure with partial sterility barriers between subpopulations. It is therefore also important to consider the genomic background in which novel or desirable alleles occur. Moreover, some alleles may have phenotypic effects that are conditional on subpopulation or some unidentified component of the genetic background. We show here a method that is capable of not only identifying population structure but also decomposing the genome of each individual into a mosaic map of segments that have origins in different subpopulations. Interestingly, this method seems to be most effective on genotype-by-sequencing data and does not require imputation. Its effectiveness and resolution on fixed array genotype data may be limited to the extent ascertainment bias masks the true diversity of the genotyped materials. Characterizing the genomes of naturally occurring highly admixed lines within core collections may identify valuable resources for quantitative trait dissection, or lines that possess broad compatibility alleles needed for development of advanced multiple parent mapping populations.