Single-step genomic prediction in small-scale populations

Implementation of genomic prediction is challenging for small-scale dairy populations. Veterinary physician Andrei Kudinov, a researcher at Natural Resources Institute Finland (LUKE), studied in his doctoral dissertation methods to implement single-step genomic prediction in small scale dairy populations. The studies were carried out at the Department of Agricultural Sciences, Natural Resources Institute Finland, and Russian Research Institute of Farm Animal Genetics and Breeding (Saint-Peterburg, Russia).

Genomic prediction is widely used in dairy cattle all around the world. The implementation of the procedure is relatively straightforward for large populations by using the two-step approach: train the model - predict breeding value. Because the quality of the prediction is relying on the size of genotyped progeny tested animals, implementation of the procedure challenging for small populations. A single-step prediction approach was proposed as a better solution for limited data however, multiple aspects are unclear.

The overall aim of Kudinov’s thesis was to identify an approach to implement single-step genomic best linear unbiased prediction in small-scale dairy cattle population. The implementation was done on two populations: the Russian Black & White and Holstein population of the Leningrad region (Russia); and the Red Dairy cattle population of Finland. Pure Leningrad region genomic prediction model included majorly genotyped cows. To improve the reliability of the genomic prediction Leningrad region data was upgraded by genotypes and phenotypes of bulls from the Nordic (Finland, Denmark, and Sweden) Holstein population. The results showed low reliability of genomic prediction explained by a small number of genotyped informative animals. Improvement of prediction in milk yield trait was observed after adding Nordic information.

One limitation of the single-step is arises from the unknown ancestry of the genotyped and non-genotyped animals in the pedigree. The perspective approach to overcome the limitation (metafounders) was implemented on Finnish Read Dairy cattle. The implementation revealed several important techniques on how to design a covariance matrix of a metafounders. Results of Kudinov’s thesis might be used by Leningrad Region farmers to improve data recording and implement genomic prediction. Research on the metafounders should be used to improve the single-step approach both in large and small populations.

Veterinary Physician Andrei Kudinov will defend his doctoral dissertation entitled "Single-step genomic prediction in small-scale populations” on 17th June 2021 at 12.00 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. 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.

The public defense can also be followed remotely using the following link: https://helsinki.zoom.us/j/66987502791 Meeting ID: 669 8750 2791

Contact: Andrei Kudinov, andrei.kudinov@helsinki.fi