Department of Agricultural, Food and Nutritional Science, University of Alberta, Edmonton, Alberta, T6G 2P5, Canada.
J Anim Sci. 2010 Jan;88(1):16-22. doi: 10.2527/jas.2008-1759. Epub 2009 Sep 11.
Feed intake and efficiency are economically important traits because feed is the greatest variable cost in beef production. Feed efficiency can be measured as residual feed intake (RFI), which is the difference between actual DMI of an animal and the expected DMI based on its BW and growth rate. Feed conversion ratio (FCR) is the inverse of gross feed efficiency and is the ratio of DMI to ADG. A total of 2,633 SNP across the 29 bovine autosomes were analyzed in 464 steers sired by Angus, Charolais, or Alberta Hybrid bulls for associations with RFI. A total of 150 SNP were associated with RFI at P < 0.05 of which 23 were significant at P < 0.01. Nine of the SNP pairs show high linkage disequilibrium (r(2) > 0.80), so only 1 of the SNP pairs was used in further multiple-marker analyses. Two methods were used to create a panel of SNP that were maximally informative for RFI based on the data. In the first method, 141 unique SNP were combined in a single multivariate model and a backward elimination model was used to drop SNP until all SNP left in the model were significant at P < 0.05. The SNP had greater effects when combined in the multivariate model than when tested individually. In the second method, the estimates from the 141 SNP were used to create a sequential molecular breeding value (MBV) according to the compound covariate prediction (CCP) procedure. The sequential MBV was built by adding the estimated effects one at a time, but only keeping SNP effects in the sequential MBV if the test statistic and the proportion of variance explained were improved. Predictabilities of the 2 methods were compared by regressing RFI on a final MBV created from SNP that remained in each analytical model. The MBV from the compound covariate prediction model produced an r(2) of 0.497, whereas the multivariate model MBV had a decreased r(2) of 0.416. The significant SNP were also tested for associations with DMI and FCR. The SNP showed different combinations of associations with the 4 traits, including some that were only associated with RFI. About 9.5% of the SNP from the 2 models were within 5 cM of previously identified RFI QTL and pinpoint areas to further explore for positional candidate genes. In conclusion, this study has identified a panel of SNP with significant effects on RFI that need to be validated in an independent population and provides continued progress toward selecting markers for use in marker-assisted selection for feed efficiency in beef cattle.
采食量和效率是具有经济重要性的特征，因为饲料是牛肉生产中最大的可变成本。饲料效率可以衡量为剩余采食量（RFI），它是动物实际采食量与根据其体重和生长速度预期采食量之间的差异。饲料转化率（FCR）是粗饲料效率的倒数，是采食量与 ADG 的比值。在 464 头由安格斯、夏洛莱或艾伯塔杂交公牛配种的公牛中，分析了 29 条牛染色体上总共 2633 个 SNP 与 RFI 的关联。共有 150 个 SNP 与 RFI 相关，其中 23 个 SNP 在 P < 0.01 时有统计学意义。9 对 SNP 之间显示出高度的连锁不平衡（r(2) > 0.80），因此仅使用 SNP 对中的 1 个 SNP 进行进一步的多标记分析。使用两种方法根据数据为 RFI 生成信息量最大的 SNP 面板。在第一种方法中，将 141 个独特的 SNP 组合在一个单一的多变量模型中，并使用向后消除模型剔除 SNP，直到模型中剩余的所有 SNP 在 P < 0.05 时有统计学意义。当 SNP 在多变量模型中组合时，其效果大于单独测试时的效果。在第二种方法中，根据复合协变量预测（CCP）程序，使用 141 个 SNP 的估计值创建顺序分子育种值（MBV）。通过一次添加估计的效应来构建顺序 MBV，但仅保留顺序 MBV 中的 SNP 效应，如果检验统计量和解释方差的比例得到改善。通过将 RFI 回归到每个分析模型中保留的 SNP 生成的最终 MBV 来比较两种方法的预测能力。来自复合协变量预测模型的 MBV 产生的 r(2)为 0.497，而多变量模型 MBV 的 r(2)降低为 0.416。还测试了与 DMI 和 FCR 相关的显著 SNP。SNP 与 4 个性状的关联表现出不同的组合，包括一些仅与 RFI 相关的 SNP。来自这两个模型的约 9.5%的 SNP 位于先前确定的 RFI QTL 内 5 cM 范围内，这为进一步探索位置候选基因提供了明确的区域。总之，本研究确定了一组对 RFI 有显著影响的 SNP，需要在独立群体中进行验证，并为选择用于牛肉牛饲料效率的标记辅助选择的标记提供了持续进展。
