Supplementary Table 1. QTL regions for each of these three sets of phenotypes from the QTL database

The application of Genomic Selection (GS) in beef cattle breeding programs is somewhat limited, despite its numerous advantages. Computer simulations are powerful tools to enhance our understanding of GS applications across different scenarios and are invaluable as an initial step before implementing this technique in "real" breeding programs. In this study, we utilized real SNP data from Indicus and Taurus breeds to simulate three crossbred breeding schemes: F1 crosses, grading up, and rotational crosses. Phenotypes were selected for shear force, growth, and tolerance traits. We compared the predictive accuracy of three 50k SNP chips that differed in SNP selection methodologies: random selection, selection based on minimum allele frequency differences between breeds, and selection based on minimum allele frequency differences between breeds with a threshold of 0.09 in Taurus. Our findings indicate that rotational crossing demonstrates optimal predictive accuracy, while selecting markers based on allele frequency differences between breeds does not yield significant benefits and may even be detrimental.

尽管基因组选择（Genomic Selection, GS）具备诸多优势，但其在肉牛育种项目中的应用仍存在一定局限。计算机模拟是帮助我们深入理解不同场景下GS应用的有力工具，同时也是在“真实”育种项目中落地该技术前的关键前置步骤，其价值不可估量。本研究使用了来自瘤牛（Indicus）和普通牛（Taurus）的真实单核苷酸多态性（Single Nucleotide Polymorphism, SNP）数据，模拟了三种杂交育种方案：F1杂交、级进杂交以及轮回杂交。本研究以剪切力、生长性状与抗逆性状为目标开展表型选择，对比了三款采用不同SNP筛选方法的50K SNP芯片的预测准确性：随机筛选、基于品种间最小等位基因频率差异的筛选，以及基于品种间最小等位基因频率差异且在普通牛群体中设置0.09阈值的筛选。研究结果显示，轮回杂交展现出最优的预测准确性；而基于品种间等位基因频率差异筛选标记的策略，不仅未带来显著收益，甚至可能产生负面影响。