August 2024 Evaluations: What’s New?

Source: CDCB

CDCB and USDA AGIL detail the changes implemented for the triannual genetic evaluations on August 13, 2024.

Introduction of 305-AA: The New standardized yield measurement

By Asha Miles, Paul VanRaden, Gary Fok, Jana Hutchison, Lillian Bacheller, Jay Megonigal and Ezequiel L. Nicolazzi

Age, parity and season adjustment factors for yield traits have been revised to replace the adjustments from 30 years ago. Standardized yield records are now adjusted to 36-months instead of mature equivalent, which was the standard from 1935 to June 2024. The revised records are referred to as 305-AA (average age of 36 months) instead of 305d-2X-ME (mature equivalent) records, or more simply, standardized records.

Effective June 12, 2024, the 305-AA yield estimates are available in CDCB’s WebConnect animal and data queries. The new 305-AA adjustments are implemented in the calculation of CDCB genetic evaluations, effective August 13, 2024.

The main goal of the new adjustments is to enable fairer phenotypic comparisons of standardized records for cows of different ages, parities and seasons of calving, based on recent analysis in the modern dairy environment.

Herd averages from standardized yields will be closer to actual yields, whereas ME records were about 10 percent higher than actual, depending on breed. Table 1 compares ME to 36-month adjustments from Schutz (1994) vs. current data for each breed.

Table 1. Ratio of mature equivalent to 36-month equivalent milk, fat and protein yields from 1994 or recent data

Predicted Transmitting Abilities (PTAs) were already revised to estimate 36-month yields at the 2005 base change. Standardized milk, fat and protein ME records were divided by 1.10 for HO and JE which was the average estimated ratio of ME to 36-month yield from Schutz (1994). Other breed ratios ranged from 1.05 to 1.15. The age factors estimated from recent data have ratios for milk closer to 1.08 of HO and JE. Fat and protein generally have smaller ratios than milk for all breeds. Those differences in standard deviation (SD) of 36-month adjusted PTA compared to previous factors will contribute to slight scaling (base) changes when implemented. The new factors are applied as a series of separate multipliers for age-parity, season, previous days open and milking frequency.

Seasonal adjustments are much less in current data than in 1994. Seasonal effects are now estimated within five regions defined by the average climate zone scores for each state. Within each region, the seasonal differences are smaller in recent decades, suggesting that improved housing and management is decreasing the effect of the environment on lactation yields. These new estimates had very little impact on PTAs because most management groups include only cows calving within the same two-month window. Separate factors were estimated for each of six five-year groups (1990, 1995, 2000, 2005, 2010, 2015).

The new factors were tested using 101.5 million milk, 100.5 million fat and 81.2 million protein lactation records from 1960 to 2022. Correlations of PTA for proven bulls born since 2000 were 0.999 for Holsteins, > 0.99 for Jerseys and Guernseys, but ranged from 0.981 to 0.984 for Brown Swiss and Milking Shorthorns. Genetic trends decreased a little for Jerseys and Brown Swiss but increased for Guernseys. Those differences in trend contributed to the lower correlations for those breeds.

Implementation of Constructed IDs across database

By Juan Pablo Nani, Lillian Bacheller, Jay Megonigal, George Wiggans, José Carrillo and Paul VanRaden

Full implementation of the Constructed ID solution across the entire CDCB database is now complete. August 2024 marks the first triannual evaluations in which this change is fully realized. The complete implementation was included in monthly and weekly evaluations starting July 2, 2024.

Constructed IDs were pioneered in the U.S. and launched in 2023 as a new solution to build more complete pedigrees for animals that previously had no link to their maternal ancestors. This results in more accurate, reliable genetic evaluations for affected animals, and in turn improves the accuracy of the U.S. evaluations overall.

Implementation of Constructed IDs began in mid-2023 for animals most immediately affected. The full “sweep” of the database was not completed until June 2024 due to the need for processing system upgrades that enabled much faster implementation of Constructed IDs over all affected pedigrees. This allowed the addition of Constructed ID relatives to over 3.2 million applicable animals within the 100+ million animals in the National Cooperator Database in less than one month. Beyond those 3.2 million, a larger number of animals were directly or indirectly affected, as the U.S. population is heavily interconnected with pedigrees of animals worldwide.

USDA Animal Genomics and Improvement Laboratory (AGIL) conducted foundational research to develop parentage identification through genomics and the resulting Constructed ID methodology. Read more here: https://uscdcb.com/constructed-ids-eliminating-pedigree-dead-ends/

Heifer livability updates and heifer termination code edits

By Mahesh Neupane, Lillian Bacheller, Kristen Gaddis, Paul VanRaden and Ezequiel L. Nicolazzi

Edits and modeling were improved for Heifer Livability (HLV). A revision in the storage of cow termination codes had accidentally prevented the most recent heifer deaths from being used in the last two years of data (2022-24). Those reports were loaded into the database.

The HLV edits and model were also revised to require at least 1 percent death loss in each year for a herd’s HLV data to be used, whereas previously the 1 percent edit was applied to a herd’s incidence across all years instead of within each year. Those revisions and corrections will greatly improve HLV evaluations for bulls whose daughters were in the last two years of data and genomic predictions for progeny of those bulls. The within-year edits will also adapt more rapidly to changes in reporting when previous herds discontinue, or new herds begin to report calf deaths in the DHI data flow.

Brown Swiss Rear Teat Placement

By Ezequiel L. Nicolazzi and Kristen Parker Gaddis

The traditional and genomic evaluation for Brown Swiss Rear Teat Placement (RTP) will be first published in August 2024. In January 2024, nearly 15,000 appraisals were used to estimate the parameters for this trait and run a first evaluation. Of these appraisals, nearly 80 percent are between 25 and 35 points on the 50-point linear scale. The heritability estimate for RTP was 0.21, in line with similar traits, such as Front Teat Placement at 0.22 heritability. Repeatability for RTP was 0.33.

Traditional evaluations – those produced from only performance data – include RTP as a single trait, currently uncorrelated to other traits.  These values will then be used to provide an RTP evaluation to all 72,000-plus genotyped Brown Swiss animals.

Ranges for Rear Teat Placement in Brown Swiss

These evaluations will be sent to Interbull in December 2024, as international validation is required to submit records internationally.

Previous SNPs removed and new gene tests added

By Dan Null, Heather A. Enzenauer, Sajjad Toghiani, Jeff O’Connell, Ahmed Al-Khudhair, Ezequiel L. Nicolazzi and Paul VanRaden

The set of 78,964 genetic markers, gene tests and quantitative trait loci used in genomic predictions since August 2021 was updated to a new list of 69,200 SNPs, effective in August 2024. The 9,769 markers removed had small effects on all traits, were not on recent chips, or had poorer genotype quality. This revision had very limited impact on evaluations. Removing some extra SNPs improved processing time, which is beneficial given the ongoing, large additions of animals in the database each year.

All SNPs on the X chromosome were retained because the X still has fewer SNPs compared to its length (X is second longest chromosome in cattle). The SNP edits considered overall call rate and minor allele frequency before removing SNPs with the smallest 20 genetic effects in each 5-Mbase chromosome window. Each SNP’s genetic standard deviation (SD) was calculated from its largest effect across all traits and breeds, thus preserving SNPs that affected any trait of any breed. The SNPs removed were 74 percent from the high-density (HD) chip and 26 percent from less dense chips such as the 50K that had low overall call rate, low minor allele frequencies, and small effects on traits. A 12 percent reduction in processing time and storage was realized, allowing focus on the most impactful and new markers.

Along with the markers removed, five new gene tests were added, including HH7 and Slick for monitoring. In a test run based on February 2024 evaluation, correlations of genomic predictions from the previous and new SNP list were about 99.6% for nearly all traits and with very little impact on averages and SDs. Animals with low density genotypes, incomplete pedigree or loose links to the reference population will be the most affected.

The SNP list revision may affect carrier status for a slightly higher percentage of animals, because some haplotypes now include a different set of markers.

Annual Breed Base Representation (BBR) Update

By Ezequiel L. Nicolazzi                                       

This year’s Breed Base Representation (BBR) update will include the usual updated set of BBR reference animals, is occurring in August, and is based on the new set of 69,200 SNPs. A test run based on the February 2024 evaluation showed comparable results to previous yearly BBR changes, confirming the little effect of the decluttered SNP list on results.

Inclusion of HH6 and JNS gene tests in haplotype calculations

By Dan Null, Heather A. Enzenauer, Lillian Bacheller, Ezequiel L. Nicolazzi and Paul VanRaden

Holstein HH6 and Jersey Neuropathy with Splayed Forelimbs (JNS) will now include direct gene test information in their haplotype calculation. Neogen regularly shares gene tests in their genotype submissions with CDCB for their use in haplotype calculation. Also, the American Jersey Cattle Association has been submitting gene test results for JNS from multiple providers to CDCB. Both these sources of information will now be included directly in imputation and will be used in haplotype calculation. Test run results show improved performance and higher correspondence for both haplotypes on animals with gene test information and their progeny.