Evaluation for the effect of low-coverage sequencing on genomic selection in large yellow croaker

Abstract

Low-coverage whole-genome sequencing (WGS) is a cost-effective genotyping technique. Combined with the imputation method, it can generate large numbers of SNPs and provide an opportunity for genomic selection (GS) using whole-genome SNPs to estimate genomic breeding values (GEBVs). However, it is unclear whether low-coverage WGS is effective for GS in large yellow croakers, even or in aquaculture populations more generally. In this study, 536 fish were sequenced with whole-genome sequencing at average depth of 8×. Low-coverage WGS datasets with different depth of 0.05×, 0.1×, 0.5×, 1×, and 4×, were generated by down-sampling of reads at 8×. For the real phenotype of visceral white spot disease and simulated polygenic traits with different heritability size and QTL numbers, we evaluate the effect of low-coverage WGS on prediction accuracy of GEBVs. The results indicate that the depth of 0.5× can almost acquire the same prediction accuracy as that of 8× in both real and simulated datasets. We also investigate the prediction accuracy of population relationship between training and validation sets. It is found that depth at 0.5× almost has the same accuracy as that of 8×. However, the accuracy of the closely-related population is twice higher than that of the distantly-related population. These findings suggest that low-coverage WGS is suitable for GS in large yellow croaker.