|SU, GUOSHENG - Aarhus University|
|NIELSEN, U - Knowledge Center For Agriculture|
|AAMAND, GERT PEDERSEN - Knowledge Center For Agriculture|
|GULDBRANDTSEN, BERNT - Aarhus University|
|LUND, MOGENS SANDØ - Aarhus University|
Submitted to: Journal of Dairy Science
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: 1/30/2015
Publication Date: 5/1/2015
Citation: Wiggans, G.R., Su, G., Cooper, T.A., Nielsen, U.S., Aamand, G., Guldbrandtsen, B., Lund, M., Van Raden, P.M. 2015. Short communication: Improving accuracy of Jersey genomic evaluations in the United States and Denmark by sharing reference population bulls. Journal of Dairy Science. 98(5):3508-3513.
Interpretive Summary: Accuracy of genomic prediction depends on how much information the reference population contributes. The gain in prediction accuracy for Jersey genomic evaluations in Denmark and the United States from using larger reference populations was assessed. Reliability of Danish genomic predictions was 4 percentage points higher with a joint reference population. For US genomic predictions, the average increase in reliability was 1 percentage point for young US animals and 2 percentage points when young Danish bulls were included in the prediction test. Exchanging reference data to increase the size of a reference population is an efficient approach to increasing the accuracy of genomic prediction.
Technical Abstract: The effect on prediction accuracy for Jersey genomic evaluations in Denmark and the United States from using larger reference populations was assessed. Each country contributed genotypes from 1,157 Jersey bulls to the reference population of the other. Eight of 9 traits analyzed by Denmark (milk, fat, and protein yields; fertility index; mastitis index; body conformation; feet-and-legs; and udder conformation) benefited from the inclusion of US Jersey bulls in the Danish reference population. Reliability gains for genomic estimated breeding value (GEBV) ranged from 0.01 for fertility to 0.13 for udder conformation. An exception was longevity, for which use of the combined Danish–US reference population resulted in a loss of 0.05 for GEBV reliability. Averaged over all 9 traits, reliability for GEBV based on the combined Danish–US reference population was 0.04 higher than reliability of GEBV based on only the Danish reference population. For the United States, mean increase in reliability over 23 yield and type traits of young US animals was 0.01 for genomic predicted transmitting ability (GPTA) based on a joint US–Danish reference population compared with previous GPTA from North American data; the reliability increase was 0.02 when young Danish bulls also were included in the validation set, which indicated a slightly larger benefit for bulls that had more Danish ancestors. To assess the effect of size of the reference population on reliability gain, the US reference population was reduced to a size similar to the Danish reference population. With the reference population subset, mean reliability gain was higher for both US and Danish young animals, which indicated a greater benefit from additional information with a smaller starting population. Exchanging reference data to increase the size of reference population was an efficient approach to increasing the accuracy of genomic prediction, especially for a breed with a small number of reference bulls.