Progress 04/01/17 to 03/31/22
Outputs
Target Audience:Animal breeding and genetics industry, animal breeding and genetics scientists and students. Changes/Problems:In the initial stages of the project, methods to predict accuracy of genomic selection using the variance of relationships between training and validation, which were derived in the literature, were found not to be valid for the types of populations relevant to animal breeding. This required substantial additional investigation by theoretical derivations and stochastic simulation, which was not anticipated at the start of theproject, requiring most of the focus in the project to be on objective 1. In addition, the post-doctoral fellow hired on the project was recruited into industry and a suitable replacement could not be secured, in part because of visa and Covid restrictions. Covid has also prevented Dr.Julius van der Werfto come to ISU to work on objective 3 or to work on objective 3 in Australia. Nevertheless, the methods and software developed will allow industry to use the results of this project for comparison of alternative breeding programs and, thereby, optimization of breeding programs. What opportunities for training and professional development has the project provided?Three post-doctoral fellows and one graduate student worked on the project and gained novel training and professional experiences. One post-doc was hired by the US animal breeding industry as a geneticist, one as an assistant professor by a US Landgrant University, and one is still a post-doctoral fellow at Iowa State University. The graduate student was hired by a plant breeding company in the US. How have the results been disseminated to communities of interest?Five presentations at international scientific conferences, two scientific journal papers, and software that will be made available to academia and industry as an R-Shiny App. What do you plan to do during the next reporting period to accomplish the goals?
Nothing Reported
Impacts
What was accomplished under these goals?
Design and optimization of modern breeding programs require methods and software to predict response to multiple-trait and multiple-stage breeding programs based on estimated breeding values and genomic predictions. Inclusion of genomic predictions in these methods requires availability of methods to predict the accuracy of genomic predictions. A new method to deterministically predict the accuracy of genomic prediction among selection candidates was developed and validated. The prediction is based on the accuracy of genomic predictions within the training data, the separation of this accuracy into that based on pedigree and genomic information, the erosion of the accuracy of each from the training to the selection candidates, and the recombining of pedigree and genomic information into the accuracy of genomicpredictions among selection candidates. In addition, a method was developed to estimate the effective number of chromosome segments (Me) in the training population based on the observed accuracy of genomic predictions in the training population. Me is needed to model the erosion of genomic information from the training population to the selection candidates. An efficient leave-one-out crossvalidation method was developed to obtain the accuracy of genomic predictions within the training population for realistic genetic evaluation models with multiple fixed and random effects. The existing SelAction software for prediction of response to pedigree-based selection was translated into Python and expanded to allow for a more exact implementation of multiple-stage selection and for the inclusion of genomic predictions. A user interface was developed using R-Shiny. Prediction of rates of inbreeding were implemented following the SelAction implementation with inclusion of genomic predictions. This software can be used by breeders to reliably compare alternative breeding programs and for investment decisions for breeding programs that include genomic information.
Publications
- Type:
Conference Papers and Presentations
Status:
Awaiting Publication
Year Published:
2022
Citation:
Dekkers, J.C.M. and A. Wolc. 2022. Opportunities offered by genomic selection to breed for diverse production systems. World Poultry Conference. Paris, France.
- Type:
Conference Papers and Presentations
Status:
Awaiting Publication
Year Published:
2022
Citation:
Dekkers, J.C.M., H. Su, L. Kramer, and H. Yu. 2022. An approach for the design of breeding programs using genomics. 12th World Congress on Genetics Applied to Livestock Production, the Netherlands.
- Type:
Journal Articles
Status:
Published
Year Published:
2022
Citation:
Bijma, P., Dekkers, J.C.M. 2022. Predictions of the accuracy of genomic prediction: connecting R2, selection index theory, and Fisher information. Genet Sel Evol 54, 13. https://doi.org/10.1186/s12711-022-00700-2
- Type:
Journal Articles
Status:
Published
Year Published:
2021
Citation:
Dekkers, J., Su, H. and Cheng, J., 2021. Predicting the accuracy of genomic predictions. Genetics Selection Evolution, 53(1), pp.1-23. https://doi.org/10.1186/s12711-021-00647-w
- Type:
Journal Articles
Status:
Published
Year Published:
2021
Citation:
Cheng, J., Dekkers, J.C. and Fernando, R.L., 2021. Cross?validation of best linear unbiased predictions of breeding values using an efficient leave?one?out strategy. Journal of Animal Breeding and Genetics. doi.org/10.1111/jbg.12545
|
Progress 04/01/20 to 03/31/21
Outputs
Target Audience:Animal breeding and genetics industry, animal breeding and genetics scientists and students. Changes/Problems:The start of objectives 2 and 3 has been delayed because of the unanticipated additional extensive work that was required to develop the methods for objective 1. In addition, the post-doctoral fellow hired on the project was recruited into industry and a suitable replacement could not be secured, in part because of visa and Covid restrictions. Covid has also prevented Dr. Julius van der Werf to come to ISU to work on objective 3. What opportunities for training and professional development has the project provided?The graduate student working on the leave-one-out approach was able to present his work at the annual meeting of the American Association of Animal Science. How have the results been disseminated to communities of interest?Presentations at two virtual scientific conferences. What do you plan to do during the next reporting period to accomplish the goals?Objective 1). Publish the two scientific manuscripts. Finalize the softwared and develop a user interface for the software. Initiate objectives 2 and 3
Impacts
What was accomplished under these goals?
Objective 1) Develop and validate deterministic methods to predict response to selection in breeding programs using genomic selection. Two manuscripts were written and submitted; one on the method developed to predict the accuracy of genomic prediction and one on the efficient leave-one-out strategy. Additional analyses and simulations were required. Objective 2) Develop and validate methods to predict and control inbreeding using genomic selection. Because of the extra work needed for objective 1 and the delays in hiring a new post-doc because of COVID, this objective has not yet been initiated. Objective 3) Develop and apply methods to optimize breeding programs using genomic selection. Because of the extra work needed for objective 1 and the inability for Dr. Julius van der Werf to travel to ISU because of COVID, this objective has not yet been initiated.
Publications
- Type:
Conference Papers and Presentations
Status:
Published
Year Published:
2020
Citation:
Cheng J., R. Fernando, and J. C. M. Dekkers. 2020. Cross validation of best linear unbiased prediction of breeding values using an efficient leave-one-out strategy. ASAS-CSAS-WSASAS Virtual Meeting & Trade Show.
|
Progress 04/01/19 to 03/31/20
Outputs
Target Audience: Animal breeding and genetics industry, animal breeding and genetics scientists and students. Changes/Problems:The start of objectives 2 and 3 has been delayed because of the unanticipated additional extensive work that was required to develop the methods for objective 1. In addition, the post-doctoral fellow hired on the project was recruited into industry and a suitable replacement could not be secured, in part because of visa and Covid restrictions. Covid has also prevented Dr. Julius van der Werf to come to ISU to work on objective 3. What opportunities for training and professional development has the project provided?One post-doctoral fellow who was hired on the project was hired by the US animal breeding industry as a geneticist. One graduate student working on the project on a part-time basis is gaining novel training and professional experiences. How have the results been disseminated to communities of interest?An abstract was submitted to the International Quantitative Genetics Conference that was to be held in Australia in August 20202. This conference will not be virtual in November 2020 and we have been asked to give an oral presentation. A manuscript on the deterministic prediction method has been submitted to the journal Genetics, Selection, Evolution. What do you plan to do during the next reporting period to accomplish the goals? Objective 2) research on prediction of inbreeding will be initiated. Objective 3) Methods to optimize breeding programs will be further developed.
Impacts
What was accomplished under these goals?
Objective 1) Develop and validate deterministic methods to predict response to selection in breeding programs using genomic selection. The new method to deterministically predict the accuracy of genomic prediction among selection candidates has now been finalized and validated. The prediction is based on the accuracy of genomic predictions within the training data, the separation of this accuracy,in that based on pedigree and genomic information, the erosion of the accuracy of each from the training to the selection candidates, and the recombining of pedigree and genomic information into the accuracyof genomic predictions among selection candidates. In addition, a method was developed to estimate the effective number of chromosome segments (Me) in the training population based on the observed accuracy of genomic predictions in the training population. Me is needed to model the erosion of genomic information from the training population to the selection candidates.An efficient leave-one-out crossvalidation method was developed to obtain the accuracy of genomic predictions within the training population for realistic genetic evaluation models with multiple fixed and random effects. The Python program for interactive evaluation of breeding programs using deterministic predictions of genetic gain was expanded to allow for multiple-stage selection and selection across groups of individuals with different amounts of information. Objective 2) Develop and validate methods to predict and control inbreeding using genomic selection. Because of the extra work needed for objective 1, this objective has not yet been initiated. Objective 3) Develop and apply methods to optimize breeding programs using genomic selection. The methods developed under objective 1 will enable optimization of breeding programs with genomics.
Publications
- Type:
Conference Papers and Presentations
Status:
Awaiting Publication
Year Published:
2020
Citation:
Dekkers, J.C.M., J. Cheng, and H. Su. Predicting the accuracy of genomic prediction. 6th International Conference on Quantitative Genetics, November 2020.
|
Progress 04/01/18 to 03/31/19
Outputs
Target Audience: Animal breeding and genetics industry, animal breeding and genetics scientists and students. Changes/Problems:Methods to predict accuracy of genomic selection using the variance of relationships between training and validation, which were derived in the literature,were found not to be valid for the types of populations relevant to animal breeding. This has required substantial additional investigation by theoretical derivations and stochastic simulation. Although new methods have been developed, these still require further validation and this has delayed the start of the objectives 2 and 3. What opportunities for training and professional development has the project provided? The Post-doctoral fellow who was hired to work on the project is gainingnovel training and professional experiences. How have the results been disseminated to communities of interest? A poster was presented at the Gordon Conference on Quantitative Genetics and Genomics in February, 2019, proving extensive opportunities for interaction with stakeholders (animal breeders and other geneticists from both academia and industry). What do you plan to do during the next reporting period to accomplish the goals? Objective 1) programs for prediction of single- and multiple-trait selection programs with genomics will be completed and a beta version will be released. The new method for deterministic prediction of the accuracy of genomic prediction will be finalized, validated, and implemented. Objective 2) research on prediction of inbreeding will be initiated. Objective 3) Methods to optimize breeding programs will be further developed.
Impacts
What was accomplished under these goals?
Objective 1) Develop and validate deterministic methods to predict response to selection in breeding programs using genomic selection. Extensive stochastic simulations were usedto validate theoretical predictions of the accuracy of genomic predictions based on the variance of genomic relationships between the training and validation populations. These predictions were found to be accuratefor full-sib family designs but to substantially overestimate the accuracy of genomic prediction for the hierarchical full-sib half-sib family designs that are typical for livestock breeding populations. This led to substantial investigations using theoretical derivations and stochastic simulation on the validity of the use of the variance of relationships between training and validation and development of a new method for the accuracy of genomic prediction, which is based on two components: 1)the variance of relationships WITHIN the training population to predict the accuracy of genomic prediction WITHIN training; 2) the erosion of this accuracy from the training to the prediction population. Analytical predictions for the latter are currently under development and validation. In addition, selection index methods were developed to combine predictions using pedigree information with the additional information that is obtained from genomics. In the mean time, the Python program for interactive evaluation of breeding programs using deterministic predictions of genetic gain is nearly complete and ready for beta testing. Objective 2) Develop and validate methods to predict and control inbreeding using genomic selection. Not yet initiated Objective 3) Develop and apply methods to optimize breeding programs using genomic selection. Methods to analytically predict the variance of genomic relationships were developed, which will enable optimization of breeding programs and phenotyping programs with genomic information.
Publications
- Type:
Conference Papers and Presentations
Status:
Published
Year Published:
2019
Citation:
Dekkers, J.C.M, and H. Su, 2019. Deterministic predictions of accuracy of genomic predictions. Gordon Research Conferences, Italy.
|
Progress 04/01/17 to 03/31/18
Outputs
Target Audience:Animal breeding and genetics industry, animal breeding and genetics scientists and students. Changes/Problems:
Nothing Reported
What opportunities for training and professional development has the project provided?A post-doctoral fellow was hired to work on the project, providing novel training and professional experiences. How have the results been disseminated to communities of interest?An abstract was presented at the Midwest Animal Science conference in March, 2018. Input on program capabilities and interface has been obtained from several breeding organizations. What do you plan to do during the next reporting period to accomplish the goals? Objective 1) programs for prediction of single- and multiple-trait selection programs with genomics will be completed and validated. Objective 2) research on prediction of inbreeding will be initiated. Objective 3) Methods to optimize breeding programs will be initiated. Drs. Julius van der Werf (Australia) and Han Mulder (Netherlands) will come to Iowa State University in June 2018 to collaborate on these aspects of the project.
Impacts
What was accomplished under these goals?
Overall impact statement: Animals and livestock contribute 40 percent of the global value of agricultural output and contribute to the livelihoods and food security of almost a billion people worldwide. Advances in animal breeding, genetics, and genomics will facilitate a more efficient industry. Efficiencyof breeding programs is critical to improving livestock production and lowering costs for producers. This project period, we have made substantial progress on development and programming of deterministic methods to predict response to single- and multiple-stage selection in livestockbreeding programs with non-overlapping (discrete) generations. Objective 1) Develop and validate deterministic methods to predict response to selection in breeding programs using genomic selection. Different programming languages were evaluated and Python was chosen because of it's universality, portability across platforms, and extensive suite of libraries. Important components of existing software SelAction for predicting response to selection for phenotype-based breeding programs with discrete generations were translated into Python and tested. Algorithms for multiple-stage selection were improved through the use of truncated multivariate normal numerical procedures. Substantial programs has been made on the theory of methods for incorporating genomic information in deterministic response predictions, in collaboration with colleagues at the University of New England in Australia (Dr. Julius van der Werf) and Wageningen University in the Netherlands (Drs. Piter Bijma and Han Mulder). Stochastic simulation programs have been developed and are being used to validate the theoretical predictions. Objective 2) Develop and validate methods to predict and control inbreeding using genomic selection. Not yet initiated Objective 3) Develop and apply methods to optimize breeding programs using genomic selection. Not yet initiated
Publications
- Type:
Conference Papers and Presentations
Status:
Published
Year Published:
2017
Citation:
H. Su, P. Bijma, J. van der Werf, and J. C. M. DekkersSoftware Development for Deterministic Prediction of Selection Response in Livestock Breeding Programs Using Genomic Information. Midwest Animal Science Meeting, Omaha, NE
|
|