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.
Foundation Animal Effects: Difference between revisions
No edit summary |
No edit summary |
||
| (23 intermediate revisions by 2 users not shown) | |||
| Line 1: | Line 1: | ||
[[Category: Genetic Evaluation]] | [[Category: Genetic Evaluation]] | ||
Implementing [[Best Linear Unbiased Prediction | BLUP animal models]] without accounting for differences in groups of [[Glossary#F | foundation animals']] genetic merit can result in [[Prediction_Bias | biased EPDs]]. Differences in additive genetic merit between groups of foundation animals can be a result of different breeds of origin, founders of the same breed entering the data used for genetic evaluation at different periods of time, or a combination of both. The differences from entering the data at different times are due to genetic trend resulting from within breed selection. | |||
__TOC__ | __TOC__ | ||
=Additive genetic groups= | =Additive genetic groups= | ||
When additive genetic group effects are not included all foundation animals are assumed to come from a population with the same average genetic merit . <ref>Westell, R. A., R. L. Quaas, and L. D. Van Vleck. 1988. Genetic | |||
groups in an animal model. J. Dairy Sci. 71:1310.</ref><ref>Quaas, R. L. 1988. Additive genetic model with groups and relationships. J. Dairy Sci. 71:1338. </ref> Including additive genetic groups | groups in an animal model. J. Dairy Sci. 71:1310.</ref><ref>Quaas, R. L. 1988. Additive genetic model with groups and relationships. J. Dairy Sci. 71:1338. </ref> Including additive genetic groups permits the model to predict genetic differences that occur between the groups. These differences impact the [[BIF recommends the use of EPD|EPDs]] of the descendants of the foundation animals. | ||
When designating genetic groups by periods of time there is a balance between size of group and amount of time to cluster groups | When designating genetic groups by periods of time there is a balance between size of group and amount of time on which to cluster groups. Too short a period results in groups' predictions with high prediction errors. Too long a time period decreases the precision of the prediction. This problem is compounded in multi-breed evaluations where groups are designated with both breed of founder and generation. | ||
A founder animal can be assigned to more than one genetic group. The genetic group assignment is determined by the | A founder animal can be assigned to more than one genetic group in a multi-breed evaluation. The genetic group assignment is determined by the animal's parents' breed group of origin. Additionally, if one parent is known (often the dam) then the animal is partially assigned to the breed of the unknown parent or year genetic group. | ||
Additive genetic groups | Additive genetic groups can be included as fixed effects by absorbing the group equations into the inverse numerator relationship matrix. Direct simplified computational procedures have been well described. While computationally easy to set up, including additive genetic groups may cause instability during iteration for the solution of the mixed model equations. More sophisticated preconditioners can be useful to improve performance to convergence during iteration. When implemented this way the resulting breeding value predictions include the fixed group effects. | ||
Alternatively, an equivalent approach is to include additive genetic groups as fixed covariates. Doing so will improve iterative performance. The covariates will be a set of very dense equations (many non-zero values). However, modern computers with long vector processing capability can process these large and dense problems effortlessly. The resulting breeding value predictions will not include the fixed group effects. These effects must be added back to the predictions so that they include all the additive genetic variance accounted for by the model. | |||
Fitting groups as fixed effects results in a non-singular set of equations. An appropriate constraint to the linear system of mixed-model equations must be applied. | |||
=Metafounders= | =Metafounders= | ||
The use of genomic data allows for the inclusion of an alternative to additive genetic groups. Unlike additive genetic groups, metafounder effects can account for relationships between foundation groups even though they may be distinct breeds.<ref>Legarra A, Christensen | The use of genomic data allows for the inclusion of an alternative to additive genetic groups. Unlike additive genetic groups, metafounder effects can account for relationships between foundation groups even though they may be distinct breeds.<ref>Legarra A., O. F. Christensen, Z. G. Vitezica, I. Aguilar, I. Misztal. 2015. Ancestral Relationships Using Metafounders: Finite Ancestral Populations and Across Population Relationships. Genetics. 2015 Jun;200(2):455-68. doi: 10.1534/genetics.115.177014. Epub 2015 Apr 14. PMID: 25873631; PMCID: PMC4492372.</ref> Because all breeds ultimately come from the same origins they share many of the same alleles. | ||
[[Genomic Data | Genomic data]] can be used to estimate the alleles that are common by descent from an original ancestral population. The metafounder approach uses a "pseudo-individual" to represent an animal of ancestral origin analogous to additive genetic groups. These pseudo individuals' relationships are estimated using pedigree and genomic data across populations.<ref>Legarra, A., M. Bermann, Q. Mei, O. F. Christensen. 2024. Redefining and interpreting genomic relationships of metafounders. Genet Sel Evol 56, 34. https://doi.org/10.1186/s12711-024-00891-w</ref> The pseudo individual then contributes to the relationship matrix calculated from the pedigree data used in the [[Multiple Trait Evaluation|mixed model equations.]] | |||
= Recommendations = | |||
''BIF Recommends: 1) If foundation animals are from different breeds of origin or believed to differ in genetic merit BIF recommends fitting of additive genetic groups, metafounders, or covariates to account for these differences; 2) Selection decisions should be taken on the combined genetic merit of individuals (EPD and group effect estimates).'' | |||
=References= | =References= | ||
Latest revision as of 22:47, 24 June 2024
Implementing BLUP animal models without accounting for differences in groups of foundation animals' genetic merit can result in biased EPDs. Differences in additive genetic merit between groups of foundation animals can be a result of different breeds of origin, founders of the same breed entering the data used for genetic evaluation at different periods of time, or a combination of both. The differences from entering the data at different times are due to genetic trend resulting from within breed selection.
Additive genetic groups
When additive genetic group effects are not included all foundation animals are assumed to come from a population with the same average genetic merit . [1][2] Including additive genetic groups permits the model to predict genetic differences that occur between the groups. These differences impact the EPDs of the descendants of the foundation animals.
When designating genetic groups by periods of time there is a balance between size of group and amount of time on which to cluster groups. Too short a period results in groups' predictions with high prediction errors. Too long a time period decreases the precision of the prediction. This problem is compounded in multi-breed evaluations where groups are designated with both breed of founder and generation.
A founder animal can be assigned to more than one genetic group in a multi-breed evaluation. The genetic group assignment is determined by the animal's parents' breed group of origin. Additionally, if one parent is known (often the dam) then the animal is partially assigned to the breed of the unknown parent or year genetic group.
Additive genetic groups can be included as fixed effects by absorbing the group equations into the inverse numerator relationship matrix. Direct simplified computational procedures have been well described. While computationally easy to set up, including additive genetic groups may cause instability during iteration for the solution of the mixed model equations. More sophisticated preconditioners can be useful to improve performance to convergence during iteration. When implemented this way the resulting breeding value predictions include the fixed group effects.
Alternatively, an equivalent approach is to include additive genetic groups as fixed covariates. Doing so will improve iterative performance. The covariates will be a set of very dense equations (many non-zero values). However, modern computers with long vector processing capability can process these large and dense problems effortlessly. The resulting breeding value predictions will not include the fixed group effects. These effects must be added back to the predictions so that they include all the additive genetic variance accounted for by the model.
Fitting groups as fixed effects results in a non-singular set of equations. An appropriate constraint to the linear system of mixed-model equations must be applied.
Metafounders
The use of genomic data allows for the inclusion of an alternative to additive genetic groups. Unlike additive genetic groups, metafounder effects can account for relationships between foundation groups even though they may be distinct breeds.[3] Because all breeds ultimately come from the same origins they share many of the same alleles.
Genomic data can be used to estimate the alleles that are common by descent from an original ancestral population. The metafounder approach uses a "pseudo-individual" to represent an animal of ancestral origin analogous to additive genetic groups. These pseudo individuals' relationships are estimated using pedigree and genomic data across populations.[4] The pseudo individual then contributes to the relationship matrix calculated from the pedigree data used in the mixed model equations.
Recommendations
BIF Recommends: 1) If foundation animals are from different breeds of origin or believed to differ in genetic merit BIF recommends fitting of additive genetic groups, metafounders, or covariates to account for these differences; 2) Selection decisions should be taken on the combined genetic merit of individuals (EPD and group effect estimates).
References
- ↑ Westell, R. A., R. L. Quaas, and L. D. Van Vleck. 1988. Genetic groups in an animal model. J. Dairy Sci. 71:1310.
- ↑ Quaas, R. L. 1988. Additive genetic model with groups and relationships. J. Dairy Sci. 71:1338.
- ↑ Legarra A., O. F. Christensen, Z. G. Vitezica, I. Aguilar, I. Misztal. 2015. Ancestral Relationships Using Metafounders: Finite Ancestral Populations and Across Population Relationships. Genetics. 2015 Jun;200(2):455-68. doi: 10.1534/genetics.115.177014. Epub 2015 Apr 14. PMID: 25873631; PMCID: PMC4492372.
- ↑ Legarra, A., M. Bermann, Q. Mei, O. F. Christensen. 2024. Redefining and interpreting genomic relationships of metafounders. Genet Sel Evol 56, 34. https://doi.org/10.1186/s12711-024-00891-w
