Skip to main content
ARS Home » Research » Publications at this Location » Publication #264713

Title: Genomic imputation and evaluation using 342 high density Holstein genotypes

item Vanraden, Paul
item Null, Daniel
item Wiggans, George
item Sonstegard, Tad
item Connor, Erin

Submitted to: Journal of Dairy Science
Publication Type: Abstract Only
Publication Acceptance Date: 2/24/2011
Publication Date: 6/30/2011
Citation: Van Raden, P.M., Null, D.J., Wiggans, G.R., Sonstegard, T.S., Connor, E.E. 2011. Genomic imputation and evaluation using 342 high density Holstein genotypes. Journal of Animal Science 89(E-Suppl. 1)/ Journal of Dairy Science 94(E-Suppl. 1):533(abstr. W53).

Interpretive Summary:

Technical Abstract: Genomic evaluations for 73,749 Holsteins were computed using 636,967 of the 777,000 markers on the Illumina high density (HD) chip. Observed data included 342 animals with HD genotypes, 54,676 animals with 42,503 marker (50K) genotypes, 17,371 animals with 2,614 marker (3K) genotypes, and 1,360 nongenotyped dams (0K) with > 90% of haplotypes imputable from progeny. The HD genotypes were from 180 influential sires, 138 Beltsville research cows, and 24 other females. Percentages of correctly imputed genotypes were estimated using an example simulated chromosome for this same population structure with 1% of genotypes missing and 0.02% incorrect initially from each chip. Over all animals, 94.4% of genotypes were missing initially. After imputation of missing markers, 99.9% of genotypes were correct from HD, 98.0% from 50K, 88.4% from 3K, and 93.8% from 0K genotypes. These imputation rates with version 2 of program findhap.f90 are several percent higher than with version 1. Version 2 begins with long segments to improve haplotype matches for close relatives and ends with short segments to detect matches from more remote ancestors instead of choosing just one optimal segment length. Evaluations were tested using imputation of actual genotypes and August 2007 phenotypes to predict deregressed evaluations of bulls proven after August 2007. For 29 traits tested, estimated genomic reliability averaged 54.3% using 636,967 markers vs. 54.8% using 42,503 regressions vs. 29.9% from traditional parent average. Squared correlations with future data were higher for 10 traits and lower for 19 with HD than with 50K evaluations. The largest marker effects were located at very similar positions, but new markers from the HD chip often had larger effects than the best markers from the 50K chip. Results were less favorable than the 1.6% increase in reliability expected from simulation, but more animals with HD genotypes will improve imputation and reliability. Also, multi-breed evaluation could produce larger gains than the single-breed evaluation investigated here.