当前位置: X-MOL 学术Genet. Sel. Evol. › 论文详情
Our official English website, www.x-mol.net, welcomes your feedback! (Note: you will need to create a separate account there.)
Predictions of the accuracy of genomic prediction: connecting R2, selection index theory, and Fisher information
Genetics Selection Evolution ( IF 4.1 ) Pub Date : 2022-02-14 , DOI: 10.1186/s12711-022-00700-2
Piter Bijma 1 , Jack C M Dekkers 2
Affiliation  

Deterministic predictions of the accuracy of genomic estimated breeding values (GEBV) when combining information sources have been developed based on selection index theory (SIT) and on Fisher information (FI). These two approaches have resulted in slightly different results when considering the combination of pedigree and genomic information. Here, we clarify this apparent contradiction, both for the combination of pedigree and genomic information and for the combination of subpopulations into a joint reference population. First, we show that existing expressions for the squared accuracy of GEBV can be understood as a proportion of the variance explained. Next, we show that the apparent discrepancy that has been observed between accuracies based on SIT vs. FI originated from two sources. First, the FI referred to the genetic component that is captured by the marker genotypes, rather than the full genetic component. Second, the common SIT-based derivations did not account for the increase in the accuracy of GEBV due to a reduction of the residual variance when combining information sources. The SIT and FI approaches are equivalent when these sources are accounted for. The squared accuracy of GEBV can be understood as a proportion of the variance explained. The SIT and FI approaches for combining information for GEBV are equivalent and provide identical accuracies when the underlying assumptions are equivalent.

中文翻译:

基因组预测准确性的预测:连接R2、选择指数理论和Fisher信息

基于选择指数理论 (SIT) 和费希尔信息 (FI) 开发了组合信息源时基因组估计育种值 (GEBV) 准确性的确定性预测。在考虑谱系和基因组信息的组合时,这两种方法导致的结果略有不同。在这里,我们澄清了这一明显的矛盾,无论是对于谱系和基因组信息的组合,还是对于将亚群组合成一个联合参考群体。首先,我们表明 GEBV 平方精度的现有表达式可以理解为解释的方差的比例。接下来,我们展示了基于 SIT 与 FI 的精度之间观察到的明显差异源自两个来源。第一的,FI 指的是由标记基因型捕获的遗传成分,而不是完整的遗传成分。其次,常见的基于 SIT 的推导没有考虑到 GEBV 准确性的提高,这是由于在组合信息源时残差方差的减少。当考虑到这些来源时,SIT 和 FI 方法是等效的。GEBV 的平方精度可以理解为解释的方差的比例。用于组合 GEBV 信息的 SIT 和 FI 方法是等效的,并且在基本假设相同时提供相同的准确度。由于组合信息源时残差方差的减少,基于 SIT 的常见推导无法解释 GEBV 准确性的提高。当考虑到这些来源时,SIT 和 FI 方法是等效的。GEBV 的平方精度可以理解为解释的方差的比例。用于组合 GEBV 信息的 SIT 和 FI 方法是等效的,并且在基本假设相同时提供相同的准确度。由于组合信息源时残差方差的减少,基于 SIT 的常见推导无法解释 GEBV 准确性的提高。当考虑到这些来源时,SIT 和 FI 方法是等效的。GEBV 的平方精度可以理解为解释的方差的比例。用于组合 GEBV 信息的 SIT 和 FI 方法是等效的,并且在基本假设相同时提供相同的准确度。
更新日期:2022-02-15
down
wechat
bug