Skip to main content
ARS Home » Research » Publications at this Location » Publication #254818
Title: Combining different marker densities in genomic evaluation

Author
item Vanraden, Paul
item O'CONNELL, J - University Of Maryland
item Wiggans, George
item WEIGEL, K - University Of Wisconsin

Submitted to: Interbull Annual Meeting Proceedings
Publication Type: Proceedings
Publication Acceptance Date: 5/31/2010
Publication Date: 5/31/2010
Citation: Van Raden, P.M., O'Connell, J.R., Wiggans, G.R., Weigel, K.A. 2010. Combining different marker densities in genomic evaluation. Interbull Annual Meeting Proceedings. Interbull Bull. 42:113–118. 2010.

Interpretive Summary: Combining genotypes of different densities will make genomic selection more efficient and affordable by allowing breeders to choose between different genetic marker sets available for different prices. The missing genotypes in the lower density sets can be filled (imputed) from genotypes or haplotypes of relatives or from matching allele patterns in the general population. Actual genotypes of 40,351 Holsteins, 4,064 Jerseys, and 1,455 Brown Swiss were used to compare a full set of 43,385 markers to a subset of 3,209 evenly spaced markers. Average gains in reliability for young animals using the 3,209 marker subset were 79-88% of those with the full set if imputing was used but only 61-63% without imputation. More precise estimates of reliability will allow breeders to properly balance benefits vs. costs of using different marker sets.

Technical Abstract: Accurate genomic evaluations are less costly if many animals are genotyped at less than the highest density and their missing genotypes filled using haplotypes. Mixed density files for 45,870 animals were examined by reducing half of young animal or all animal genotypes from the observed 43,385 markers to a subset of 3,209 markers. For young Holsteins genotyped with 3,209 markers, the gain in net merit reliability was 79% of the gain from genotyping 43,385 markers. When half of the reference population had 3,209 markers, gain was 90% for young animals with 43,385 markers and 73% for young animals with 3,209 markers. Gain was 66% when all animals had only 3,209 markers. Simulated gain in reliability from increasing the number of markers to 500,000 was only 1.4%, but more than half of that gain could result from genotyping just 1,586 bulls at higher density. Reliability improved when more reference animals were genotyped at higher density. Individual reliabilities can be adjusted to account for number of markers and success of imputation.