According to the doctoral dissertation of Andrei Kudinov, small cattle breeds and populations can implement genomic prediction by genotyping many cows and including data from external populations. Some finetuning of specific statistical models can be needed.
Over the last decade synergy of genomic and phenotypic data has been used by cattle farmers and breeders to find the genetically best parents of future generations. The procedure is known widely as genomic selection or genomic evaluation. Its use is relatively straightforward in populations with a large amount of data where there are many informative animals such as bulls with high prediction accuracy. However, in small populations having a low number of informative animals and a high level of germplasm importation use of the method might be challenging.
In his dissertation, Andrei Kudinov presented a way to implement genomic prediction in cattle populations with a low number of genotyped animals. He studied inclusion of genotyped animals with low information into the prediction, particularly use of genotyped cows. In addition, approaches to incorporate information from an external population were examined. The work was done at Natural Resources Institute Finland (Luke) and used data from the Leningrad region (Saint-Petersburg) dairy farms and the joint Nordic (Finland, Denmark, and Sweden) dairy cattle evaluation.
State of art genomic prediction
Accuracy of genomic prediction was low, as expected, when the phenotypic and genomic data were from the Leningrad region only. Integration of the Nordic genotypes and phenotypes improved accuracy of the prediction. This external information could be used to improve prediction by the Russian farmers, but the accumulation of local genotypes must continue and certain improvements in the phenotypic recording should be done.
“This is an excellent example of international collaboration between many Nordic countries and Russia under Luke leadership”, Kudinov says.
Kudinov studied also so called metafounder approach in genomic prediction. This new approach was tested for the Finnish Red Dairy cattle. In this study, a subset of Finnish data was used to emulate a small population. The study was the first example of the metafounder method on a non-simulated data.
“Application of theory is not always a trivial task. It was also the case here. Nordic Red Dairy cattle has a complicated genetic structure, like many dairy breeds, with an incomplete pedigree”, Kudinov says.
The developed ways to implement the metafouder approach on real data was already used as the source of knowledge in Nordic dairy cattle evaluations.
Veterinary Physician Andrei Kudinov will defend his doctoral dissertation entitled “Single-step genomic prediction in small-scale populations“ on 17th June 2021 at 14:15 in the Faculty of Agriculture and Forestry, University of Helsinki.
The public defense will take place in lecture room 115, Kielikeskus (Language Centre), Fabianinkatu 26, Helsinki. Link to the remote connection public defence for the audience: https://helsinki.zoom.us/j/66987502791.
Alternatively, you can join the same Zoom event using the following information: Webinar ID: 669 8750 2791
Doctor Ole Christensen from Center for Quantitative Genetics and Genomics, Aarhus University, Denmark will serve as the opponent and Professor Pekka Uimari as the custos. The dissertation is published in the series Dissertationes Schola Doctoralis Scientiae Circumiectalis, Alimentariae, Biologicae.