Limits...
GplusE: beyond genomic selection

View Article: PubMed Central - PubMed

ABSTRACT

GplusE is a strategy for genomic selection in which the accuracy of assessment in the reference population for a primary trait such as yield is increased by the incorporation of data from high‐ throughput field phenotyping platforms. This increase in precision comes from both exploiting genetic relationships between traits and reducing the effect of environmental influences upon them. We describe a collaborative project among researchers and breeders to develop a large reference population of elite UK wheat lines. This will be used to test the method, to study the design of the reference population, and to test genotyping strategies and imputation methods. Finally, it will provide data to pump‐prime the application of genomic selection to UK winter wheat breeding.

No MeSH data available.


Related in: MedlinePlus

Partial pedigree of 44 crosses selected for GplusE project. Crosses are represented by “X”. Upper panel: descendents of the variety Moulin (listed 1984), the most recent common ancestor of 43 out of 44 of the selected crosses. Lower panel: ancestors of an arbitrarily selected cross. The positions of Moulin and Robigus, a recent ancestor of many of the lines on the current UK recommended list, are identified. Plots were created using Pedigree Viewer (Kinghorn 1994).
© Copyright Policy - creativeCommonsBy
Related In: Results  -  Collection

License
getmorefigures.php?uid=PMC4998132&req=5

fes352-fig-0001: Partial pedigree of 44 crosses selected for GplusE project. Crosses are represented by “X”. Upper panel: descendents of the variety Moulin (listed 1984), the most recent common ancestor of 43 out of 44 of the selected crosses. Lower panel: ancestors of an arbitrarily selected cross. The positions of Moulin and Robigus, a recent ancestor of many of the lines on the current UK recommended list, are identified. Plots were created using Pedigree Viewer (Kinghorn 1994).

Mentions: In our study, with the collaboration of four breeders – Elsoms, KWS, Limagrain and RAGT – we are creating a large population of at least 3000 wheat lines from up to 44 elite crosses. Representative views of the linked pedigree are given in Figure 1, and illustrate the consanguineous, unstructured and intermeshed nature of the UK winter wheat pedigree. Allele frequencies at genetic markers have changed over time in the United Kingdom (White et al. 2008), but genetic variation has not declined; it increases in periods when more independent breeding programs contribute varieties. The 44 crosses have been selected to cover the diversity of parents that breeders are currently using. The lines we generate will be phenotyped over 2 years and genotyped with the 35 k UK Affymetrix SNP chip (http://www.affymetrix.com/). The large size of the population, and the known pedigree structure, will allow its partition into RPs and test populations with varying degrees of relationship between the two to assess how accuracy of prediction varies with genetic distance and marker density. This approach has been simulated for wheat by Hickey et al. (2014) and the results of those simulations have been used in the experimental design of GplusE. For example, at one extreme, the reference and training populations can be constructed so that for every cross, there are representatives in both. Alternatively, the two populations could be selected to minimize the pedigree relationship between them. This might involve ensuring that they contain no common grandparents or more distant relationships. In between these extremes are partitions where, say, if two crosses have a single parent in common, then lines from one are allocated to the test population only and lines from the other to the reference population only, while individuals from the same cross are never represented in both. There are many other possible partitions, and selection can also take place using marker‐based estimates of kinship rather than those from the pedigree. Superimposed on these alternatives, the size of the reference population can be varied. In comparing results from these alternatives, we hope to understand how to construct the most effective RP for the germplasm which UK breeders are currently using.


GplusE: beyond genomic selection
Partial pedigree of 44 crosses selected for GplusE project. Crosses are represented by “X”. Upper panel: descendents of the variety Moulin (listed 1984), the most recent common ancestor of 43 out of 44 of the selected crosses. Lower panel: ancestors of an arbitrarily selected cross. The positions of Moulin and Robigus, a recent ancestor of many of the lines on the current UK recommended list, are identified. Plots were created using Pedigree Viewer (Kinghorn 1994).
© Copyright Policy - creativeCommonsBy
Related In: Results  -  Collection

License
Show All Figures
getmorefigures.php?uid=PMC4998132&req=5

fes352-fig-0001: Partial pedigree of 44 crosses selected for GplusE project. Crosses are represented by “X”. Upper panel: descendents of the variety Moulin (listed 1984), the most recent common ancestor of 43 out of 44 of the selected crosses. Lower panel: ancestors of an arbitrarily selected cross. The positions of Moulin and Robigus, a recent ancestor of many of the lines on the current UK recommended list, are identified. Plots were created using Pedigree Viewer (Kinghorn 1994).
Mentions: In our study, with the collaboration of four breeders – Elsoms, KWS, Limagrain and RAGT – we are creating a large population of at least 3000 wheat lines from up to 44 elite crosses. Representative views of the linked pedigree are given in Figure 1, and illustrate the consanguineous, unstructured and intermeshed nature of the UK winter wheat pedigree. Allele frequencies at genetic markers have changed over time in the United Kingdom (White et al. 2008), but genetic variation has not declined; it increases in periods when more independent breeding programs contribute varieties. The 44 crosses have been selected to cover the diversity of parents that breeders are currently using. The lines we generate will be phenotyped over 2 years and genotyped with the 35 k UK Affymetrix SNP chip (http://www.affymetrix.com/). The large size of the population, and the known pedigree structure, will allow its partition into RPs and test populations with varying degrees of relationship between the two to assess how accuracy of prediction varies with genetic distance and marker density. This approach has been simulated for wheat by Hickey et al. (2014) and the results of those simulations have been used in the experimental design of GplusE. For example, at one extreme, the reference and training populations can be constructed so that for every cross, there are representatives in both. Alternatively, the two populations could be selected to minimize the pedigree relationship between them. This might involve ensuring that they contain no common grandparents or more distant relationships. In between these extremes are partitions where, say, if two crosses have a single parent in common, then lines from one are allocated to the test population only and lines from the other to the reference population only, while individuals from the same cross are never represented in both. There are many other possible partitions, and selection can also take place using marker‐based estimates of kinship rather than those from the pedigree. Superimposed on these alternatives, the size of the reference population can be varied. In comparing results from these alternatives, we hope to understand how to construct the most effective RP for the germplasm which UK breeders are currently using.

View Article: PubMed Central - PubMed

ABSTRACT

GplusE is a strategy for genomic selection in which the accuracy of assessment in the reference population for a primary trait such as yield is increased by the incorporation of data from high‐ throughput field phenotyping platforms. This increase in precision comes from both exploiting genetic relationships between traits and reducing the effect of environmental influences upon them. We describe a collaborative project among researchers and breeders to develop a large reference population of elite UK wheat lines. This will be used to test the method, to study the design of the reference population, and to test genotyping strategies and imputation methods. Finally, it will provide data to pump‐prime the application of genomic selection to UK winter wheat breeding.

No MeSH data available.


Related in: MedlinePlus