Changes to evaluation system (April 2014)
Improved discovery of maternal grandsires
By George Wiggans and Lillian Bacheller
The pedigree maternal grandsire (MGS) is checked for consistency as each genotype is processed. If the pedigree MGS is determined to be unlikely or unknown, all bulls that are enough older than the animal to be an MGS and not one of the excluded relatives are considered as a possible MGS. A procedure that checks 1 single-nucleotide polymorphism (SNP) at a time has been used. A recent report by van Kaam and Hayes (2013) suggested that greater accuracy could be achieved if whole intervals (half the chromosome plus 1 SNP) were checked as a unit. A test set of 5,133 genotypes was selected at random from families in which the animal, sire, dam, and MGS were all genotyped with more than 3,000 SNP. The single-SNP method found the true MGS 92.6% of the time, which was 1.2 percentage points more often than the interval method; therefore, the single-SNP method was retained. Based on analysis of true MGS, the single-SNP method was revised to allow MGS with a conflict percentage that caused them to be classified as unlikely to be proposed as the discovered MGS. The threshold for conflict percentage for a bull to be considered as an MGS was increased by 3 percentage points, and the minimum age of an MGS at the animal’s birth was reduced to 1,150 days. In addition, more bulls were considered after initial screening based on 1,000 SNP. Those changes increased the accuracy in the test set by 5.5 percentage points.
Reference:
van Kaam, J.B.C.H.M., and B.J. Hayes. 2013. Maternal grandsire verification and detection without imputation. Interbull Bull. 47:120–124.
Revised LD chip and incorporation of gene tests
By George Wiggans, Paul VanRaden, and Dan Null
The low density (LD) chip from Illumina was revised slightly; the new version will be referred to as LD2 with chip number 13 in formats 38 and 105. Chips LD and LD2 are nearly synonymous, with changes much smaller than between the 50K chip versions 1 and 2. Compared with LD, the LD2 excludes 5 previous markers and includes 8 additional markers from the 50K chip. Numbers of markers currently used are 6,785 from LD and 6,787 from LD2.
Several tests for simply inherited genetic conditions that previously were available on the GeneSeek Genomic Profiler version 2 (GP2) and GeneSeek High Density (GHD) for individual animals are now provided to CDCB for all animals. These tests include Arachnomelia, Beta-Lactoglobulin, Bovine Growth Hormone Receptor, BLAD, Citrullinemia, DUMPS, Leptin, Recessive Red, Mulefoot haplotype, SDM, SMA, and Weaver haplotype. Incorporation of direct gene tests or targeted markers for more animals will improve the haplotype tests for animals previously genotyped with chips that did not contain these and will allow more conditions to be reported and genomic predictions to be improved.
Revised calculation of genomic future inbreeding
By Chuanyu Sun and Paul VanRaden
Genomic future inbreeding (GFI) for Holsteins is now calculated as the average genomic relationship of each animal to proven bulls born in the last 10 years, whereas previously and for the other breeds the definition includes proven bulls and cows with records born in the last 10 years. In April, 73,422 Holstein cows had records in the reference population. Time required to compute genomic relationships for 400,637 young animals with all of the reference animals became excessive, whereas calculation of relationships with the 25,011 reference bulls was more manageable (< 1 day). The slightly revised definition caused only very small changes to GFI. Another purpose of the genomic relationships is to compare to pedigree relationships in order to detect incorrect pedigrees of imputed dams, but some of these may no longer be detected because of the reduced computation.
Jersey genotypes from Denmark
The genotype exchange with Denmark was previously announced in January, and monthly evaluations since then have included the additional Danish genotypes. This is just a reminder that April is the first full release when the added genotypes affect all bulls that were previously evaluated in December. For further information see: