Project : USDA ARS

ARS Home » Northeast Area » Beltsville, Maryland (BARC) » Beltsville Agricultural Research Center » Animal Genomics and Improvement Laboratory » Research » Research Project #442491

Research Project: Increasing Accuracy of Genomic Prediction, Developing Algorithms, Selecting Markers, and Evaluating New Traits to Improve Dairy Cattle

Location: Animal Genomics and Improvement Laboratory

2024 Annual Report

Objectives
Objective 1: Expand the data used in genomic prediction by selecting new variants from five dairy breeds that more precisely track the true gene mutations that cause phenotypic differences. The goal in Objective 1 is to use high-quality sequence data from multiple dairy breeds to identify new variants for inclusion in genomic predictions, making them more representative of all dairy animals, and improving their reliability compared to currently used single-nucleotide polymorphism markers in genotyping arrays. Sub-objective 1.A: Validate genomic predictions for U.S. cows. Sub-objective 1.B: Track haplotypes to identify new, large-effect QTL for inclusion in genomic evaluations. Sub-objective 1.C: Interrogate whole genome sequences (WGS) from multiple dairy breeds to characterize their genetic differences and select high-impact variants. Objective 2: Evaluate new traits that can all be predicted at birth from the same inexpensive DNA sample. Sub-objective 2.A: Identify new traits or data types via analysis of industry trends and determine their suitability for inclusion in genomic evaluations. Sub-objective 2.B: Develop new traits to promote animal health and welfare by creating new data pipelines and eventual GPTA. Sub-objective 2.C: Update NM$ with new traits and changing prices. Objective 3: Improve efficiency of genomic prediction and computation by developing faster algorithms with better statistical properties. Sub-objective 3.A: Expand the use of international phenotypes to improve the accuracy of prediction because U.S. bulls are the most frequently used worldwide. Sub-objective 3.B: Develop and test methods to select for heat tolerance and efficiency in variable climates by examining the genotype by environment interaction. Sub-objective 3.C: Update lactation record adjustment factors and yield projection factors used in predictive models. Sub-objective 3.D: Account for genomic pre-selection biases by comparing multi-step and single-step evaluation methods towards the goal of applying single-step to routine national evaluations.

Approach
Obj. 1: Genomic predictions will be validated using > 1 million U.S. cows genotyped early in life that were phenotyped later in life. Methods to detect and impute lethal recessive alleles will continue to be improved by investigating haplotypes with no or few homozygotes, estimating if conception or stillbirth rates are affected, inspecting sequence variants to determine the most likely causal variant, and incorporating lethal recessives discovered by other researchers into the U.S. evaluation. To improve genetic progress, genetic effects will be estimated for haplotypes already defined from array genotypes. By matching haplotypes with largest effects to sequence data, better markers for net merit and individual traits will be discovered. Documenting how well genomic predictions correspond to cow performance will increase breeder confidence and participation in national evaluation systems. Obj. 2: New data on Johne’s disease, milking speed, milk mid-infrared spectra, hoof health, beef x dairy inseminations, and beef x dairy calving ease will be explored to develop further management or selection tools. High-throughput phenotyping and big data analytics will be explored, industry trends documented, and economic values estimated for genetic evaluation for current and new traits to update the national selection index. Edits will be revised to accept data from herds that record milk yield but without approved fat or protein component testing. Such herds will add to reliability for traits with less data or lower heritability even if their additional records add little to the already high reliability of genomic prediction for yield traits. Embryo transfer (ET) has grown exponentially in the last few years but has not yet been reflected in reporting breeding records. To prevent bias resulting from ET calves with missing implantation breeding events or erroneous AI events, program alterations for unreported ET will be explored for reproductive performance evaluations (e.g., conception rates, gestation length, early first calving, and daughter pregnancy rate) and improved data pipelines for ET reporting will be developed. Obj. 3: Foreign phenotypes from more countries will be incorporated into U.S. genomic predictions such as for feed intake. Differences in temperature and humidity across states and seasons will be used to rank bulls for their daughter production in hotter or cooler regions. Careful choice of scale for reporting the interactions and extra education will be needed to help breeders understand the new rankings. Models for evaluating crossbred animals will be refined. Age adjustment factors and yield projection factors used in predictive models will be updated to account for genetic and management changes in recent decades. Multi-step and single-step evaluation methods will be compared in cooperation with the University of Georgia to better account for genomic pre-selection biases with the goal to apply single-step to routine national evaluations.

Progress Report
Progress was successfully made on all three of the key Objectives of project 8042-31000-113-000D. Under Objective 1 (Expand the data used in genomic prediction by selecting new variants), genomic predictions were validated for five U.S. dairy breeds and 17 economic traits by ARS scientists at Beltsville, Maryland, for genotyped heifers that later were phenotyped as cows, and the results were submitted for publication. Revised methods to validate genomic predictions for bulls were also tested in cooperation with the Council on Dairy Cattle Breeding and Lactanet. A novel algorithm using FST approach for prioritizing influential variant within QTL regions was tested successfully in simulation of high-density SNP data in collaboration with University of Georgia; further testing the algorithm when numbers and positions of QTL are unknown is warranted and under investigation. ARS scientists took the lead detecting a new mutation that results in calf muscle weakness, and results are now published scientifically and for all 5 million genotyped Holsteins to help owners reduce the incidence of this unfavorable condition using the direct genetic test and mating programs. AGIL scientists’ collaborative work with university partners resulted in publications describing first ever million-cow genome-wide associated study for different production and fertility traits in U.S. Holstein cows, and another publication detailing the discovery of lethal haplotypes in Nellore cattle. QTLs from these studies can be used in genomic prediction, leading to improved animal health, fertility, and production. Under Objective 2 (Evaluate new traits), an ARS researcher took the lead in collaboration with industry and university partners to develop milking speed as a new trait in the U.S. genetic evaluation system. This new trait ‘milking speed’ will likely be incorporated in routine U.S. dairy genetic evaluations this year. Results on milking speed were presented at international meetings and a publication has been drafted. ARS scientists have analyzed data from millions of beef-on-dairy matings and are estimating the effects of breeding beef males to dairy females. Computer programs were developed to process the largest dataset available for this kind of study, and models for this comprehensive dataset are being developed to assess the impacts of beef service sires on reproductive performance and disease incidence in dairy cattle and calf’s health, production, and performance. This research will identify optimal strategies and quantify potential economic value of incorporating beef genetics into dairy breeding programs. Under Objective 3 (Improve efficiency of genomic prediction and computation), ARS scientists cooperated with researchers from University of Connecticut to characterize genotype by environment interaction in dairy cattle in different regions of the U.S. The research determined the impact of genotype by environmental interactions in dairy animals from California and New England for milk yield, fat and protein content, and somatic cells score. These results will help to provide resources to tackle future challenges posed by climate change through increasing the competitiveness and sustainability of dairy facilities in specific environmental regions with fewer herds and animals. ARS scientists cooperated with the Council on Dairy Cattle Breeding to test single-step evaluation methods to compute national genetic evaluation methods. Other ARS scientists’ collaborative efforts resulted in publications including improved genomic prediction in crossbred dairy cattle, estimation of inbreeding depression in U.S. Holstein bulls, new models for more precise test-day milk yields, and improved accuracy of imputation in cross breed U.S. dairy cattle. These findings will help to improve efficiency of genomic prediction and decrease computational cost.

Accomplishments
1. Effects of dairy cattle size and production on feed intake. Nutritionists previously had not estimated feed requirements for individual milk components or individual traits that affect body size and instead used only milk energy and body weight. Scientists from ARS in Beltsville, Maryland, obtained more accurate and precise predictions of feed intake and maintenance costs using 8,513 lactations of 6,621 Holstein cows in cooperation with four other U.S. dairy research herds. The scientists investigated phenotypic and genomic regressions of feed intake predicted using three components of milk (protein, butterfat, and lactose) and five predictors of body weight. The Net Merit index was revised to better account for preliminary estimates and will be revised again for the published estimates. Use of the revised indexes will make dairy cows more profitable by selecting for smaller size, improved feed efficiency, and higher milk production. Feed efficiency in the dairy cattle industry represents a major opportunity to enhance sustainability and farm profitability through genetic selection.

2. Validating the benefits from genomic selection of dairy heifers. Millions of young dairy heifers have had their DNA inspected to predict which will become the most profitable cows. ARS scientists from Beltsville, Maryland, previously developed methods to check the accuracy of genomic predictions for bulls and adapted those methods to validate heifer predictions across five U.S. dairy breeds. The goal was to determine if the predictions for each of 17 traits are accurate and unbiased by comparing genomic and parent average evaluations to later data of the same animal. The research utilized the Council on Dairy Cattle Breeding's official evaluation data from the August 2021 database to predict the corresponding trait data from the August 2023 database. This research sheds light on the advantages of using genomic predictions on a global scale. Results show that herd owners in the dairy industry may experience greater benefits from genomics than originally expected. Farmers using these accurate predictions are rapidly changing their breeding practices to improve profit.

3. Milking speed as a new trait for genetic selection. Many large farms now use in-line milk meters that automatically collect milk weights from every milking. In cooperation with North Carolina State, ARS scientists at Belstville, Maryland, characterized quantitative measurements of milking speed coming off of these systems and their suitability for genetic selection. The primary aim of this research was to identify sources of confounding bias attributable to system and biological effects as well as critical data cleaning parameters, given that this type of data was not designed for research use. This research identified breed, lactation number, and stage in lactation as biological parameters impacting milking speed phenotypes, as well as the system influences of meter manufacturer and milking frequency. Meter manufacturer had a particularly large effect in robotic milking systems. No strong association between milking speed and udder health parameters were indicated which is of large concern to dairy producers.

U.S. DEPARTMENT OF AGRICULTURE

Animal Genomics and Improvement Laboratory: Beltsville, MD