Optimizing genomic selection for scarcely recorded traits

M.J. Pszczola

Research output: Thesisinternal PhD, WUAcademic

Abstract

Animal breeding aims to genetically improve animal populations by selecting the best individuals as parents of the next generation. New traits are being introduced to breeding goals to satisfy new demands faced by livestock production. Selecting for novel traits is especially challenging when recording is laborious and expensive and large scale recording is not possible. Genetic improvement of novel traits may be thus limited due to the small number of observations. New breeding tools, such as genomic selection, are therefore needed to enable the genetic improvement of novel traits. Using the limited available data optimally may, however, require alternative approaches and methodologies than currently used for conventional breeding goal traits. The overall objective of this thesis was to investigate different options for optimizing genomic selection for scarcely recorded novel traits. The investigated options were: (1) genotype imputation for ungenotyped but phenotyped animals to be used to enlarge the reference population; (2) optimization of the design of the reference population with respect to the relationships among the animals included in it; (3) prioritizing genotyping of the reference population or the selection candidates; and (4) using easily recordable predictor traits to improve the accuracy of breeding values for scarcely recorded traits. Results showed that: (1) including ungenotyped animals to the reference population can lead to a limited increase in the breeding values accuracy; (2) the reference population is designed optimally when the relationship within the reference are minimized and between reference population and potential selection candidates maximized; (3) the main gain in accuracy when moving from traditional to genomic selection is due to genotyping the selection candidates, but preferably both reference population and selection candidates should be genotyped; and (4) including the predictor traits in the analysis when it is recorded on both reference population and selection candidates can lead to a significant increase in the selection accuracy. The key factors for successful implementation of selection for a novel trait in a breeding scheme are: (1) maximizing accuracy of genotype prediction for ungenotyped animals to be used for updating the reference population; (2) optimizing the design of the reference population; (3) determining easy to record indicator traits that are also available on the selection candidates (4) developing large scale phenotyping techniques; and (5)  establishing strategies and policies for increasing the engagement of farmers in the recording of novel traits.

Original languageEnglish
QualificationDoctor of Philosophy
Awarding Institution
  • Wageningen University
Supervisors/Advisors
  • van Arendonk, Johan, Promotor
  • Calus, Mario, Co-promotor
  • Strabel, T., Co-promotor, External person
Award date22 Nov 2013
Place of PublicationWageningen
Publisher
Print ISBNs9789461737663
Publication statusPublished - 2013

Fingerprint

marker-assisted selection
breeding
animals
breeding value
genotyping
genetic improvement
genotype
animal breeding
livestock production
farmers
phenotype
prediction

Keywords

  • dairy cattle
  • genomes
  • selective breeding
  • genetic improvement
  • breeding value
  • phenotypes
  • genotypes
  • traits
  • feed intake
  • animal breeding

Cite this

Pszczola, M. J. (2013). Optimizing genomic selection for scarcely recorded traits. Wageningen: Wageningen UR.
Pszczola, M.J.. / Optimizing genomic selection for scarcely recorded traits. Wageningen : Wageningen UR, 2013. 158 p.
@phdthesis{7ca81b5c726a45e2acb3cb29eb9ec08a,
title = "Optimizing genomic selection for scarcely recorded traits",
abstract = "Animal breeding aims to genetically improve animal populations by selecting the best individuals as parents of the next generation. New traits are being introduced to breeding goals to satisfy new demands faced by livestock production. Selecting for novel traits is especially challenging when recording is laborious and expensive and large scale recording is not possible. Genetic improvement of novel traits may be thus limited due to the small number of observations. New breeding tools, such as genomic selection, are therefore needed to enable the genetic improvement of novel traits. Using the limited available data optimally may, however, require alternative approaches and methodologies than currently used for conventional breeding goal traits. The overall objective of this thesis was to investigate different options for optimizing genomic selection for scarcely recorded novel traits. The investigated options were: (1) genotype imputation for ungenotyped but phenotyped animals to be used to enlarge the reference population; (2) optimization of the design of the reference population with respect to the relationships among the animals included in it; (3) prioritizing genotyping of the reference population or the selection candidates; and (4) using easily recordable predictor traits to improve the accuracy of breeding values for scarcely recorded traits. Results showed that: (1) including ungenotyped animals to the reference population can lead to a limited increase in the breeding values accuracy; (2) the reference population is designed optimally when the relationship within the reference are minimized and between reference population and potential selection candidates maximized; (3) the main gain in accuracy when moving from traditional to genomic selection is due to genotyping the selection candidates, but preferably both reference population and selection candidates should be genotyped; and (4) including the predictor traits in the analysis when it is recorded on both reference population and selection candidates can lead to a significant increase in the selection accuracy. The key factors for successful implementation of selection for a novel trait in a breeding scheme are: (1) maximizing accuracy of genotype prediction for ungenotyped animals to be used for updating the reference population; (2) optimizing the design of the reference population; (3) determining easy to record indicator traits that are also available on the selection candidates (4) developing large scale phenotyping techniques; and (5)  establishing strategies and policies for increasing the engagement of farmers in the recording of novel traits.",
keywords = "melkvee, genomen, selectief fokken, genetische verbetering, fokwaarde, fenotypen, genotypen, kenmerken, voeropname, dierveredeling, dairy cattle, genomes, selective breeding, genetic improvement, breeding value, phenotypes, genotypes, traits, feed intake, animal breeding",
author = "M.J. Pszczola",
note = "WU thesis 5614",
year = "2013",
language = "English",
isbn = "9789461737663",
publisher = "Wageningen UR",
school = "Wageningen University",

}

Pszczola, MJ 2013, 'Optimizing genomic selection for scarcely recorded traits', Doctor of Philosophy, Wageningen University, Wageningen.

Optimizing genomic selection for scarcely recorded traits. / Pszczola, M.J.

Wageningen : Wageningen UR, 2013. 158 p.

Research output: Thesisinternal PhD, WUAcademic

TY - THES

T1 - Optimizing genomic selection for scarcely recorded traits

AU - Pszczola, M.J.

N1 - WU thesis 5614

PY - 2013

Y1 - 2013

N2 - Animal breeding aims to genetically improve animal populations by selecting the best individuals as parents of the next generation. New traits are being introduced to breeding goals to satisfy new demands faced by livestock production. Selecting for novel traits is especially challenging when recording is laborious and expensive and large scale recording is not possible. Genetic improvement of novel traits may be thus limited due to the small number of observations. New breeding tools, such as genomic selection, are therefore needed to enable the genetic improvement of novel traits. Using the limited available data optimally may, however, require alternative approaches and methodologies than currently used for conventional breeding goal traits. The overall objective of this thesis was to investigate different options for optimizing genomic selection for scarcely recorded novel traits. The investigated options were: (1) genotype imputation for ungenotyped but phenotyped animals to be used to enlarge the reference population; (2) optimization of the design of the reference population with respect to the relationships among the animals included in it; (3) prioritizing genotyping of the reference population or the selection candidates; and (4) using easily recordable predictor traits to improve the accuracy of breeding values for scarcely recorded traits. Results showed that: (1) including ungenotyped animals to the reference population can lead to a limited increase in the breeding values accuracy; (2) the reference population is designed optimally when the relationship within the reference are minimized and between reference population and potential selection candidates maximized; (3) the main gain in accuracy when moving from traditional to genomic selection is due to genotyping the selection candidates, but preferably both reference population and selection candidates should be genotyped; and (4) including the predictor traits in the analysis when it is recorded on both reference population and selection candidates can lead to a significant increase in the selection accuracy. The key factors for successful implementation of selection for a novel trait in a breeding scheme are: (1) maximizing accuracy of genotype prediction for ungenotyped animals to be used for updating the reference population; (2) optimizing the design of the reference population; (3) determining easy to record indicator traits that are also available on the selection candidates (4) developing large scale phenotyping techniques; and (5)  establishing strategies and policies for increasing the engagement of farmers in the recording of novel traits.

AB - Animal breeding aims to genetically improve animal populations by selecting the best individuals as parents of the next generation. New traits are being introduced to breeding goals to satisfy new demands faced by livestock production. Selecting for novel traits is especially challenging when recording is laborious and expensive and large scale recording is not possible. Genetic improvement of novel traits may be thus limited due to the small number of observations. New breeding tools, such as genomic selection, are therefore needed to enable the genetic improvement of novel traits. Using the limited available data optimally may, however, require alternative approaches and methodologies than currently used for conventional breeding goal traits. The overall objective of this thesis was to investigate different options for optimizing genomic selection for scarcely recorded novel traits. The investigated options were: (1) genotype imputation for ungenotyped but phenotyped animals to be used to enlarge the reference population; (2) optimization of the design of the reference population with respect to the relationships among the animals included in it; (3) prioritizing genotyping of the reference population or the selection candidates; and (4) using easily recordable predictor traits to improve the accuracy of breeding values for scarcely recorded traits. Results showed that: (1) including ungenotyped animals to the reference population can lead to a limited increase in the breeding values accuracy; (2) the reference population is designed optimally when the relationship within the reference are minimized and between reference population and potential selection candidates maximized; (3) the main gain in accuracy when moving from traditional to genomic selection is due to genotyping the selection candidates, but preferably both reference population and selection candidates should be genotyped; and (4) including the predictor traits in the analysis when it is recorded on both reference population and selection candidates can lead to a significant increase in the selection accuracy. The key factors for successful implementation of selection for a novel trait in a breeding scheme are: (1) maximizing accuracy of genotype prediction for ungenotyped animals to be used for updating the reference population; (2) optimizing the design of the reference population; (3) determining easy to record indicator traits that are also available on the selection candidates (4) developing large scale phenotyping techniques; and (5)  establishing strategies and policies for increasing the engagement of farmers in the recording of novel traits.

KW - melkvee

KW - genomen

KW - selectief fokken

KW - genetische verbetering

KW - fokwaarde

KW - fenotypen

KW - genotypen

KW - kenmerken

KW - voeropname

KW - dierveredeling

KW - dairy cattle

KW - genomes

KW - selective breeding

KW - genetic improvement

KW - breeding value

KW - phenotypes

KW - genotypes

KW - traits

KW - feed intake

KW - animal breeding

M3 - internal PhD, WU

SN - 9789461737663

PB - Wageningen UR

CY - Wageningen

ER -

Pszczola MJ. Optimizing genomic selection for scarcely recorded traits. Wageningen: Wageningen UR, 2013. 158 p.