Comprehensive comparison between structural variants and single-nucleotide polymorphism in genomic selection of rice (Oryza sativa L.)

Recent genomic research on plant populations has revealed abundant structural variations (SVs) of > 50 bp within and among genotypes of major crop species, but the application of SVs as a unique type of molecular marker in genetic research and breeding remains limited. In this study, we used a panel of 305 rice accessions from the 3000 Rice Genome Project to compare 26,000 + SVs (> 90% deletions and translocations) and 180,000 + high-quality SNPs in predicting the field performance of 20 traits using nine commonly used genomic selection models. We demonstrated that SVs had slightly lower prediction accuracies (PA) across 20 different traits than SNPs, but saved 53.8%–77.8% of computation time to achieve reasonably high PA. The major advantage of SVs as valuable markers in obtaining reliable genomic estimated breeding values (GEBVs) in GS experiments resulted primarily from their greater phenotypic effects on high heritability traits. Our results suggested that P_0.01 was an appropriate threshold for marker filtering of both SVs and SNPs to achieve high PA in genomic selection experiments. A dataset containing all identified SVs from this study has been deposited at the China National Center for Bioinformation under BioProject: PRJCA038102, a publicly accessible molecular marker resource, enabling broad applications in rice research including gene/QTL discovery, allelic mining, and genomic selection experiments.