In aquaculture, the proportion of edible meat (FY = fillet yield) is of major economic importance, and breeding animals of superior genetic merit for this trait can improve efficiency and profitability. Achieving genetic gains for fillet yield is possible using a pedigree-based best linear unbiased prediction (PBLUP) model with direct and indirect selection. To investigate the feasibility of using genomic selection (GS) to improve FY and body weight (BW) in rainbow trout, the prediction accuracy of GS models was compared to that of PBLUP. In addition, a genome-wide association study (GWAS) was conducted to identify quantitative trait loci (QTL) for the traits. All analyses were performed using a two-trait model with FY and BW, and variance components, heritability, and genetic correlations were estimated without genomic information. The data used included 14,165 fish in the pedigree, of which 2742 and 12,890 had FY and BW phenotypic records, respectively, and 2484 had genotypes from the 57K single nucleotide polymorphism (SNP) array. The heritabilities were moderate, at 0.41 and 0.33 for FY and BW, respectively. Both traits were lowly but positively correlated (genetic correlation; r = 0.24), which suggests potential favourable correlated genetic gains. GS models increased prediction accuracy compared to PBLUP by up to 50% for FY and 44% for BW. Evaluations were found to be biased when validation was performed on future performances but not when it was performed on future genomic estimated breeding values. The low but positive genetic correlation between fillet yield and body weight indicates that some improvement in fillet yield may be achieved through indirect selection for body weight. Genomic information increases the prediction accuracy of breeding values and is an important tool to accelerate genetic progress for fillet yield and growth in the current rainbow trout population. No significant QTL were found for either trait, indicating that both traits are polygenic, and that marker-assisted selection will not be helpful to improve these traits in this population.

中文翻译：

---

基因组选择模型使用多性状模型和多代后代测试显着提高了虹鳟鱼片产量和体重遗传价值预测的准确性

在水产养殖中，可食用肉的比例（FY = 鱼片产量）具有重要的经济意义，为此性状培育具有优良遗传价值的动物可以提高效率和盈利能力。使用具有直接和间接选择的基于谱系的最佳线性无偏预测 (PBLUP) 模型，可以实现鱼片产量的遗传增益。为了研究使用基因组选择 (GS) 来提高虹鳟鱼的 FY 和体重 (BW) 的可行性，将 GS 模型的预测准确性与 PBLUP 的预测准确性进行了比较。此外，还进行了全基因组关联研究 (GWAS) 以确定这些性状的数量性状位点 (QTL)。所有分析均使用具有 FY 和 BW 的双性状模型进行，并且在没有基因组信息的情况下估计方差分量、遗传力和遗传相关性。使用的数据包括谱系中的 14,165 条鱼，其中 2742 条和 12,890 条分别具有 FY 和 BW 表型记录，2484 条具有来自 57K 单核苷酸多态性 (SNP) 阵列的基因型。FY 和 BW 的遗传力适中，分别为 0.41 和 0.33。这两个性状均呈低但正相关（遗传相关性；r = 0.24），这表明潜在的有利相关遗传增益。与 PBLUP 相比，GS 模型将 FY 的预测准确度提高了 50%，BW 的预测准确度提高了 44%。当对未来的表现进行验证时，发现评估存在偏差，但对未来的基因组估计育种值进行验证时，则不会。鱼片产量和体重之间的低但正的遗传相关性表明通过间接选择体重可以实现鱼片产量的一些提高。基因组信息提高了育种值的预测准确性，是加速当前虹鳟鱼鱼片产量和生长遗传进展的重要工具。这两个性状都没有发现显着的 QTL，表明这两个性状都是多基因的，标记辅助选择无助于改善该群体中的这些性状。