Enhanced comment feature has been enabled for all readers including those not logged in. Click on the Discussion tab (top left) to add or reply to discussions.
Accuracy: Difference between revisions
No edit summary |
|||
| (22 intermediate revisions by 2 users not shown) | |||
| Line 1: | Line 1: | ||
Information used to calculate EPDs for a particular animal can include pedigree information, own performance data, genomic data, | Information used to calculate [[Expected Progeny Difference | EPDs]] for a particular animal can include pedigree information, own performance data, genomic data, information on descendants, and correlated traits. Accuracy depends on the value (amount and quality) of the information used for EPD calculation. | ||
Beef cattle EPDs in | Beef cattle EPDs in North America report the BIF Accuracy along with EPDs. However, in theoretical animal breeding and in genetic evaluations in many other countries and species, a different measure of accuracy, the correlation between the estimated breeding value (EBV; =2 x EPD) and the true breeding value (BV), is used. Therefore, it is important to be aware of the difference. | ||
===BIF Accuracy=== | ===BIF Accuracy=== | ||
BIF accuracy was developed to facilitate users' understanding and utilization of prediction accuracy. It is calculated as, | BIF accuracy was developed to facilitate users' understanding and utilization of prediction accuracy. It is calculated as, | ||
| Line 9: | Line 8: | ||
</math> | </math> | ||
</center> | </center> | ||
===Accuracy and reliability=== | |||
The correlation between EBV and the true BV, <math>r_{EBV,BV}</math> is, | The correlation between EBV and the true BV, <math>r_{EBV,BV}</math> is, | ||
<center><math> | <center><math> | ||
| Line 19: | Line 19: | ||
</math></center> | </math></center> | ||
and, | and, | ||
<center><math>r_{EBV,BV}=\sqrt{1-\left(1-BIF\right)^{2}} | <center><math>r_{EBV,BV}=\sqrt{1-\left(1-Accuracy_{BIF}\right)^{2}} | ||
</math></center> | </math></center> | ||
Theoretically, the prediction error variance is obtained from the inverse of the coefficient matrix of the mixed model equations. However, in most applications the coefficient matrix is too large to invert and approximations are generally used. The known approximation methods | Theoretically, the prediction error variance is obtained from the inverse of the coefficient matrix of the [[Best Linear Unbiased Prediction | mixed-model equations]]. However, in most applications the coefficient matrix is too large to invert and approximations are generally used. The known approximation methods<ref>Harris, B. et al. 1998. Approximate Reliability of Genetic Evaluations Under an Animal Model. [https://www.journalofdairyscience.org/article/S0022-0302(98)75829-1/abstract?code=jods-site Journal of Dairy Science , Volume 81, Issue 10, 2723 - 2728]</ref><ref>Meyer, K. 1989. Approximate accuracy of genetic evaluation under an animal model. [https://www.livestockscience.com/article/0301-6226(89)90041-9/abstract Livestock Production Science , Volume 21, Issue 2, 87 - 100]</ref><ref>Misztal, I. et al. 2013. Methods to approximate reliabilities in single-step genomic evaluation. [https://www.journalofdairyscience.org/article/S0022-0302(12)00802-8/abstract Journal of Dairy Science , Volume 96, Issue 1, 647 - 654]</ref> may overestimate the accuracy, especially for young animals and for the contribution made by correlated traits. An alternative that has been implemented is to estimate the prediction error variance from the posterior variance obtained from an MCMC sampler (e.g., Gibb's sampler). | ||
<ref>Harris, | |||
[https://www.journalofdairyscience.org/article/S0022-0302(98)75829-1/abstract?code=jods-site Journal of Dairy Science , Volume 81 , Issue 10 , 2723 - 2728] | Most organizations provide a table of [[Possible Change | possible change values]] that relate BIF accuracy to the expected standard error of prediction of EPDs. | ||
</ref> | ===References=== | ||
<ref>Meyer, K. 1989. Approximate accuracy of genetic evaluation under an animal model | [[Category: Genetic Evaluation]] | ||
[https://www.livestockscience.com/article/0301-6226(89)90041-9/abstract Livestock Production Science , Volume 21 , Issue 2 , 87 - 100]</ref><ref>Misztal, I. et al. 2013. Methods to approximate reliabilities in single-step genomic evaluation | |||
[https://www.journalofdairyscience.org/article/S0022-0302(12)00802-8/abstract Journal of Dairy Science , Volume 96 , Issue 1 , 647 - 654]</ref> | |||
Latest revision as of 12:45, 4 July 2024
Information used to calculate EPDs for a particular animal can include pedigree information, own performance data, genomic data, information on descendants, and correlated traits. Accuracy depends on the value (amount and quality) of the information used for EPD calculation. Beef cattle EPDs in North America report the BIF Accuracy along with EPDs. However, in theoretical animal breeding and in genetic evaluations in many other countries and species, a different measure of accuracy, the correlation between the estimated breeding value (EBV; =2 x EPD) and the true breeding value (BV), is used. Therefore, it is important to be aware of the difference.
BIF Accuracy
BIF accuracy was developed to facilitate users' understanding and utilization of prediction accuracy. It is calculated as,
Accuracy and reliability
The correlation between EBV and the true BV, is,
Of course, the additive genetic variance in the denominator terms above should be multiplied by one plus the inbreeding coefficient for each animal. The relationship between the two measures of accuracy is,
and,
Theoretically, the prediction error variance is obtained from the inverse of the coefficient matrix of the mixed-model equations. However, in most applications the coefficient matrix is too large to invert and approximations are generally used. The known approximation methods[1][2][3] may overestimate the accuracy, especially for young animals and for the contribution made by correlated traits. An alternative that has been implemented is to estimate the prediction error variance from the posterior variance obtained from an MCMC sampler (e.g., Gibb's sampler).
Most organizations provide a table of possible change values that relate BIF accuracy to the expected standard error of prediction of EPDs.
References
- ↑ Harris, B. et al. 1998. Approximate Reliability of Genetic Evaluations Under an Animal Model. Journal of Dairy Science , Volume 81, Issue 10, 2723 - 2728
- ↑ Meyer, K. 1989. Approximate accuracy of genetic evaluation under an animal model. Livestock Production Science , Volume 21, Issue 2, 87 - 100
- ↑ Misztal, I. et al. 2013. Methods to approximate reliabilities in single-step genomic evaluation. Journal of Dairy Science , Volume 96, Issue 1, 647 - 654
