APRIL CHANGES
Genomic evaluation of Guernseys
By Tabatha Cooper, George Wiggans,
Sophie Eaglen, Janez Jenko, William Luff, John Woolliams, and Brian
Schnebly
Genotypes from 2,376
Guernsey bulls and cows from collaboration between the United States, Canada,
the United Kingdom, and the Isle of Guernsey are the basis for launching the
U.S. genomic evaluation of Guernsey cattle. A study in which data from August 2011
were used to predict April 2015 performance showed a gain in reliability over
parent averages of 16.8 percentage points averaged across traits. Breed
determination uses 21 markers that are nearly monomorphic (over 90%) in Guernseys and have less than 30% of animals homozygous for
that allele in Holsteins, Jerseys, Brown Swiss, and Ayrshires. The number of
markers is small because finding ones that meet the requirements becomes more
difficult as more breeds are added. A major genetic effect was discovered on
chromosome 19 near 27,000,000 base pairs. Its effect is as large as that of the
gene for diacylglycerol O-acyltransferase 1 (DGAT1) and affects many
more traits (milk, productive life, somatic cell score, daughter pregnancy
rate, cow conception rate, size, rump, udder, and teat length). However, it
does not affect NM$; therefore, the trait effects must be almost canceling. The
new Guernsey evaluations will be provided in the same formats and on the same
schedule as for the other breeds.
The research leading to
these results has received funding from the European Union's Seventh Framework Programme for research, technological development and
demonstration under grant agreement no 289592 -Gene2Farm. Select Sires (Plain
City, OH) and the American Guernsey Association (Columbus, OH) contributed to
project development and provided genotypes; CDCB (Bowie, MD) supplied pedigree,
performance, and genotypic data.
Breed base representation for crossbreds
By Paul VanRaden,
Tabatha Cooper, Jay Megonigal, Duane Norman, and João Dürr
Most
crossbreds have not been included in genomic evaluations because marker effects
are computed separately within breeds. Edits that determine which animals are
evaluated use a small set of breed-check markers. Using all markers allows each
animal's ancestry to be estimated more precisely. Breed base representation
(BBR) will now estimate the percentage of DNA contributed to the animal by each
of 5 evaluated breeds: Holstein, Jersey, Brown Swiss, Ayrshire, and Guernsey.
These 5 new fields sum to 100 (with a minimum of 0 and a maximum of 100). BBR
values of 94 to 99% are set to 100% such values occur often even for animals
with 100% purebred ancestry. The initial BBR estimates have a standard error of
about 2% caused by normal variation within a breed as well as additional error
caused by imputation from lower density chips. BBR values will be distributed
only once for each animal, and update files will then include only the new
animals.
The genotyped,
progeny-tested bulls within each breed of evaluation serve as the reference
population for that breed. Scandinavian Red bulls are included in the Ayrshire
population and are all treated as if purebred Ayrshire. The BBR values can
provide (1) information about breed composition that is more accurate and much
easier to interpret than breed-check markers and (2) a method for combining the
marker effects from different breeds into accurate genomic predicted
transmitting abilities (GPTAs) for crossbreds. Such GPTAs must be computed on
the all-breed instead of within-breed bases, and crossbred GPTAs for
conformation traits are difficult for that reason. About 12,000 crossbred
animals were not evaluated previously. It is believed that the use of BBR can
provide the means for making genetic predictions for crossbreds possible in the
future. For further information, see:
· VanRaden,
P.M., and T.A. Cooper. 2015. Genomic
evaluations and breed composition for crossbred U.S. dairy cattle. Interbull Bull. 49:19–23. | PowerPoint
presentation
Edits and adjustments for heifer conception rate
By Jana Hutchison, Paul VanRaden, and Leigh Walton
Age limits and age groups were
updated to include heifer conception rate (HCR) records for younger animals.
Previously, records had been excluded for heifers inseminated before 1 year of
age. New edit limits include heifers inseminated at 8
months of age and older, and another age group was added in HCR for records
from the youngest animals. About 297,231 records were added for HCR, which is
about 3% of total records. Use of records from these younger animals should
improve timeliness and reliability of HCR evaluations. Earlier fertility is
desired in recent years because of greater use of sexed semen and sires with
improved calving ease as well as earlier maturity either from improved
management or genetic differences.
The HCR model previously
used a constant across all years to adjust for reduced fertility of sexed
semen. The new model estimates within-year differences between conception rates
(CRs) based on conventional or sexed-semen breedings,
and those estimates are:
|
Year |
2006 |
2007 |
2008 |
2009 |
2010 |
2011 |
2012 |
2013 |
2014 |
2015 |
|
CR
difference (%) |
25.8 |
14.1 |
18.1 |
16.4 |
15.1 |
15.1 |
17.0 |
15.0 |
12.4 |
9.5 |
Conception rates for sexed semen have greatly
improved in the last 2 years (from more than 15 percentage points lower than
conventional semen in past years to less than 10 percentage points lower in
2015). This better fertility is likely the result of improved sorting methods
and products with reduced sort ratios. The new adjustment factors will be
implemented along with the age edit change for HCR. Sexed-semen adjustments
will also be revised for cow conception rate, but differences are smaller than
for heifers because conception rates are lower and affect fewer records because
of less use of sexed semen for cows.
Mutations in HCD and in BH2
By Dan Null and Paul VanRaden
The Holstein haplotype test for
cholesterol deficiency (HCD) was improved by using the exact location of the
mutation. Two research groups (Charlier,
2016; Menzi et al., 2016)
reported that the mutation is a mobile element insertion of DNA from another
chromosome into an exon of gene APOB at location 77,958,994 on chromosome 11
(UMD3). Previously animals were labeled as carriers only if they received the
full haplotype of length 3.5 Mbase from bull Storm,
but now only the portion of the haplotype containing the mutation is required.
Previously 32,712 animals were code 1 carriers with pedigree verification, and
5,643 additional animals became code 1 using the known location of the
mutation. Similarly, 27,658 were code 3 possible carriers without pedigree
versification, and 5,704 code 3 animals were added. Soon, direct test results
could also be included within the haplotype to further improve accuracy, as is
done with several other recessive haplotypes.
Brown Swiss haplotype 2 (BH2) test
results were also improved using the exact location of the mutation at
11,063,520 on chromosome 19 (Schwarzenbacher et
al., 2016). A 1.1 Mbase region containing
the mutation had previously been used to examine crossover haplotypes, and 49
additional carriers were identified using the exact location. Direct laboratory
tests for the BH2 mutation in gene TUBD1 may be available in the near future.
Nearly all calves homozygous for BH2 or for HCD die at young ages. Genomic
testing, selection, and mating programs are all useful to reduce the occurrence
of these and other recessive defects.
Reliability and inbreeding in weekly evaluations
By Paul VanRaden,
Jay Megonigal, and Gary Fok
Genomic reliability (GREL), genomic inbreeding, and genomic future inbreeding (GFI)
have been provided in the weekly automated processing since January 2016,
whereas previously those fields were computed only during the monthly
reprocessing of all data. Weekly evaluations include only animals with new
genotypes or pedigrees that changed (Wiggans
et al., 2015), and an
approximate 2-part instead of 3-part selection index is used to compute GPTA. A
similar 2-part instead of 3-part selection index approximation was developed to
compute GREL. To compute GFI, the relationship of each animal to an average
genotype for reference bulls born in the last 10 years was computed instead of
computing all individual relationships and then averaging. To compute expected
future inbreeding (EFI), the reference bulls and their ancestors were included
in the pedigree file; for each new animal, an average pedigree
relationship to the reference bulls was computed using the method of Colleau (2002)
and software provided by Ignacio Aguilar and Ignacy Misztal (University of
Georgia, Athens, GA).
Results for a test of
weekly data were consistent with the following official monthly evaluation for
11,426 animals that were in both. For Holstein net merit, GREL of the new
animals averaged 72.2% with a standard deviation (SD) of 2.2 percentage points
for the weekly evaluation compared with 72.4% (SD of 2.2) for the monthly full
evaluation; GREL correlation for the weekly and monthly evaluations was
>0.99. New animal GFI averaged 6.9% (SD of 0.9) for the weekly data compared
with 6.7% (SD of 0.8) for the monthly data; correlation of weekly and monthly
GFIs was 0.98. The average EFI was 4.9% (SD of 1.5) for weekly data compared
with 5.5% (SD of 1.7) for monthly data. The animals' own genomic inbreeding
averaged 5.9% (SD of 3.5) for both weekly and monthly data (correlation of
1.0), whereas own pedigree inbreeding averaged 4.7% for weekly compared with
4.5% for monthly data (correlation of 0.96). Pedigree corrections during the
week after genotypes arrived caused some of those differences.
Good approximations for
GREL and inbreeding fields were obtained by outputting summary data for the
reference bulls during monthly processing and inputting that summary data
during weekly processing. Full monthly processing now takes more than 4 days;
if sufficiently accurate, the weekly system could replace some of the monthly
evaluations as computing times continue to increase.
References
·
Colleau,
J.J. 2002. An indirect approach to the extensive calculation of
relationship coefficients. Genet. Sel. Evol.
34:409–421.
·
Wiggans, G.R.,
P.M. VanRaden, and T.A. Cooper. 2015. Technical note: Rapid calculation of genomic
evaluations for new animals. J. Dairy Sci. 98:2039–2042.