Prediction of genomic breeding values based on pre-selected SNPs using ssGBLUP, WssGBLUP and BayesB for Edwardsiellosis resistance in Japanese flounder

Abstract

Background: Edwardsiella tarda causes acute symptoms with ascites in Japanese founder (Paralichthys olivaceus) and is a major problem for China’s aquaculture sector. Genomic selection (GS) has been widely adopted in breeding industries because it shortens generation intervals and results in the selection of individuals that have great breeding potential with high accuracy. Based on an artifcial challenge test and re-sequenced data of 1099 founders, the aims of this study were to estimate the genetic parameters of resistance to E. tarda in Japanese founder and to evaluate the accuracy of single-step GBLUP (ssGBLUP), weighted ssGBLUP (WssGBLUP), and BayesB for improving resistance to E. tarda by using three subsets of pre-selected single nucleotide polymorphisms (SNPs). In addition, SNPs that are associated with this trait were identifed using a single-SNP genome-wide association study (GWAS) and WssGBLUP. Results: We estimated a heritability of 0.13±0.02 for resistance to E. tarda in Japanese founder. One million SNPs at fxed intervals were selected from 4,978,724 SNPs that passed quality controls. GWAS identifed signifcant SNPs on chromosomes 14 and 24. WssGBLUP revealed that the putative quantitative trait loci on chromosomes 1 and 14 contained SNPs that explained more than 1% of the genetic variance. Three 50 k-SNP subsets were pre-selected based on diferent criteria. Compared with pedigree-based prediction (ABLUP), the three genomic methods evaluated resulted in at least 7.7% greater accuracy of predictions. The accuracy of these genomic prediction methods was almost unchanged when pre-selected trait-related SNPs were used for prediction. Conclusions: Resistance to E. tarda in Japanese founder has a low heritability. GWAS and WssGBLUP revealed that the genetic architecture of this trait is polygenic. Genomic prediction of breeding values performed better than ABLUP. It is feasible to implement genomic selection to increase resistance to E. tarda in Japanese founder with 50 k SNPs. Based on the criteria used here, pre-selection of SNPs was not benefcial and other criteria for pre-selection should be considered.