The predictive ability of indirect genetic models is reduced when culled animals are omitted from the data
The predictive ability of indirect genetic models is reduced when culled animals are omitted from the data
复制标题
DOI：
10.1186/s12711-020-0527-x
复制
发表时间：
2020-02-10
影响因子：
4.1
通讯作者：
Nielsen, Hanne M.
中科院分区：
文献类型：
--
作者：
Ask, Birgitte;Christensen, Ole F.;Nielsen, Hanne M.
研究背景从群体中剔除个体会导致群体规模的减少和群体组成的变化，这可能会影响动物对群体配偶表型的间接遗传效应的预测能力。我们假设，在分析中包括剔除动物和无表型动物的间接遗传效应会影响遗传参数的估计，提高预测能力，并减少预测育种值的偏倚。我们通过应用不同的编辑程序（即从数据中省略个体或群体）和遗传模型（即经典和间接遗传模型（IGM），根据在围栏或空间允许中花费的时间的相对比例，不考虑或考虑间接遗传效应的权重）来对此进行测试。数据包括11，111个组中123，567头猪的平均日增重，其中25%组中3%的个体被提前从组中移除。结果无论采用何种编辑程序和IGM，总遗传力的估计值（0.29 ~ 0.34）均高于直接遗传力（0.23 ~ 0.25）。从数据中省略个人或群体的间接遗传效应估计的预测能力降低了8%至46%，直接和间接遗传效应组合估计的预测能力降低了4%。遗漏全组在预测育种值中引入了偏倚。间接遗传效应的加权降低了他们估计的预测能力至少19%，直接和间接遗传效应的估计值降低了1%。结论我们发现了对猪生长的显着间接遗传效应。剔除的动物既不应从数据中删除，也不应通过基于在围栏中花费的时间或空间允许的相对比例在模型中加权其间接遗传效应来解释，因为这会降低预测能力并增加预测育种值的偏倚。剔除动物的信息对于预测间接遗传效应非常重要，必须在IGM分析中通过纳入基于围栏内花费的相对时间或相对空间允许量的固定回归来说明。
Background Physical removal of individuals from groups causes reductions in group sizes and changes in group composition, which may affect the predictive ability of estimates of indirect genetic effects of animals on phenotypes of group mates. We hypothesized that including indirect genetic effects of culled animals and of animals without phenotypes in the analysis affects estimates of genetic parameters, improves predictive ability, and reduces bias of predicted breeding values. We tested this by applying different editing procedures, i.e. omission of individuals or groups from the data, and genetic models, i.e. a classical and an indirect genetic model (IGM) without or with weighting of indirect genetic effects based on the relative proportion of time spent in the pen or space allowance. Data consisted of average daily gain for 123,567 pigs in 11,111 groups, from which 3% of individuals in 25% of groups were prematurely removed from the group. Results The estimate of total heritability was higher (0.29 to 0.34) than that of direct heritability (0.23 to 0.25) regardless of the editing procedures and IGM used. Omission of individuals or groups from the data reduced the predictive ability of estimates of indirect genetic effects by 8 to 46%, and the predictive ability of estimates of the combined direct and indirect genetic effects by up to 4%. Omission of full groups introduced bias in predicted breeding values. Weighting of indirect genetic effects reduced the predictive ability of their estimates by at least 19% and of the estimates of the combined direct and indirect genetic effects by 1%. Conclusions We identified significant indirect genetic effects for growth in pigs. Culled animals should neither be removed from the data nor accounted for by weighting their indirect genetic effects in the model based on the relative proportion of time spent in the pen or space allowance, because it will reduce predictive ability and increase bias of predicted breeding values. Information on culled animals is important for prediction of indirect genetic effects and must be accounted for in IGM analyses by including fixed regressions based on relative time spent within the pen or relative space allowance.