Elsen Jean-Michel, Raoul Jérôme, Gilbert Hélène
GenPhySE, Université de Toulouse, INRAE, Castanet-Tolosan, France.
Institut de l'Elevage, Castanet-Tolosan, France.
Genet Sel Evol. 2025 Feb 17;57(1):5. doi: 10.1186/s12711-025-00953-7.
Gametic variances can be predicted from the outcomes of a genomic prediction for any genotyped individual. This is widely used in plant breeding, applying the utility criterion (UC). This paper aims to examine the conditions to use UC for recurrent selection in livestock. Here, the UC for a selection candidate is the linear combination of the expected value of the future progeny (half of the candidate's breeding value) and its predicted gametic variance weighted by a coefficient to be optimized.
First, generalizing previous results, we derived analytically the ratio of the variance of the candidate's gametic variance and that of half of the candidate's breeding value. This ratio depends strongly on the number of quantitative trait loci (QTL) affecting the trait and, to a lesser extent, on the distribution of QTL allele frequencies: highly unbalanced frequencies and a limited number of QTL (< 10) favor higher values of the ratio. Then, changes in average breeding values and genetic variances when recurrent selection in a population of infinite size is applied were analytically derived and analyzed for selection up to 15 generations: in this ideal situation, after 5 to 10 generations (depending on ), the expected breeding values were higher with selection on UC and the genetic variance was always higher than with selection on estimated breeding values. To describe the potential of the UC in more general situations, simulations were applied to a population of 1000 males and 1000 females, with various selection rates, numbers and allele frequencies of QTL, and . These simulations were performed assuming independent QTL with known positions and effects. The best values for (i.e. providing the best genetic progress) were generally lower than 1, limiting the weight on the gametic variance. As expected from the analytical derivations, the gain in genetic progress from using UC was greatest when there were few QTL and allele frequencies were unbalanced, but they barely exceeded 5%.
We conclude that the key factor to choose selection on UC rather than on estimated breeding values is the ratio between the variance of the gametic standard deviations and the variance of the breeding values (GEBV), which should be carefully evaluated.
对于任何已进行基因分型的个体，配子方差可根据基因组预测结果进行预测。这在植物育种中被广泛应用，并采用效用标准（UC）。本文旨在研究在畜牧中使用UC进行轮回选择的条件。在此，选择候选个体的UC是未来后代预期值（候选个体育种值的一半）与其预测配子方差的线性组合，该线性组合由一个待优化的系数加权。
首先，在推广先前结果的基础上，我们通过分析得出了候选个体配子方差与候选个体育种值一半的方差之比。该比例强烈依赖于影响该性状的数量性状位点（QTL）的数量，在较小程度上还依赖于QTL等位基因频率的分布：高度不平衡的频率和有限数量的QTL（<10）有利于该比例的较高值。然后，通过分析得出并分析了在无限大小的群体中进行轮回选择时，平均育种值和遗传方差的变化，选择代数可达15代：在这种理想情况下，经过5至10代（取决于 ），基于UC进行选择时预期育种值更高，且遗传方差始终高于基于估计育种值进行选择时的情况。为了描述UC在更一般情况下的潜力，对一个由1000头雄性和1000头雌性组成的群体进行了模拟，设置了不同的选择率、QTL的数量和等位基因频率以及 。这些模拟是在假设QTL具有已知位置和效应且相互独立的情况下进行的。 的最佳值（即提供最佳遗传进展的值）通常低于1，这限制了配子方差的权重。正如分析推导所预期的那样，当QTL数量较少且等位基因频率不平衡时，使用UC获得的遗传进展增益最大，但增益几乎不超过5%。
我们得出结论，选择基于UC而非估计育种值进行选择的关键因素是配子标准差的方差与育种值（基因组估计育种值，GEBV）的方差之比，对此应仔细评估。
