Research Project: Multi-Dimension Phenotyping to Enhance Prediction of Performance in Swine

Location: Genetics and Animal Breeding

2024 Annual Report

Objectives
Objective 1. Identify proteins, genes, and quantitative trait loci associated with important phenotypes to improve efficiency of swine production. Sub-objective 1.A: Improving methodologies to enhance the analysis of gene expression data. Sub-objective 1.B: Improve nutrient utilization and conversion of feed to muscle. Sub-objective 1.C: Improve lifetime reproductive performance. Objective 2. Develop machine learning methods and artificial intelligence models that incorporate production data, biological assays, and data acquired via electronic monitoring. Sub-objective 2.A: Enable prediction of animal performance to inform management decisions in real-time. Sub-objective 2.B: Enhance selection decisions to improve performance of future generations.

Approach
Demand for pork products continues to grow, while producers are being asked to reduce their environmental impact; therefore, commercial swine production needs to reduce inefficiencies while continuing to improve product quality and quantity. Inefficiencies are magnified by a pig’s response to stressors encountered, whether these stressors are environmental (temperature), health (pathogens) or social interactions with pen-mates. The greatest opportunities for increased efficiency are to improve nutrient utilization and increase sow lifetime reproductive performance, by reducing reproductive failure and culling associated with lameness. These challenges are complex and require novel approaches to rapidly identify solutions. The proposed research will contribute to the development of more accurate computational tools for incorporating production data, biological assays, and data acquired via electronic monitoring into selection decisions. These tools will be crucial to create a catalogue of genetic variants that alter gene function (either in amount or function of protein) and trait expression. Putative functional genetic variants identified in our research population will be used to inform selection decisions in commercial populations once their effects have been validated. Our research will focus on identification of genetic markers responsible for variation in conversion of feed to pork product and lifetime reproductive performance in commercial sows. In addition to genetic variant discovery, tools developed in the proposed research will allow early identification of sick or stressed animals via real-time monitoring through electronic systems. This will improve animal welfare and reduce associated mitigation costs. The unique swine resources at USMARC will be coupled with our genomic capabilities to permit more accurate prediction of phenotypes or breeding values, contributing to the goals of National Program 101’s Action Plan. These predictions will inform herd management decisions, resulting in improved performance in current pigs as well as enhance selection decisions to create more productive pigs in future generations.

Progress Report
Improvement of methodologies for analyzing gene expression data has been the focus of several projects (Objective 1A). In 2023, a gene classification algorithm was generated to classify genes according to their expression distribution. Porcine genes from 77 tissues, provided by collaborators from the Farm Animal Genotype-Tissue Expression (FarmGTEx; https://www.researchgate.net/project/FarmGTEx-Project) project, were classified using this algorithm. Gene classifications were used to develop a novel approach for identifying differentially expressed genes. Comparison of accuracies between the new approach and existing methods is currently being conducted. We continue to utilize automated feeding systems to study both feeding behavior and feed efficiency in pigs (Objective 1B). Currently, we possess records for more than 14,000 pigs, and data on approximately 1,900 pigs is added each year. For pigs in the grow-finish phase of production, ARS in Clay Center, Nebraska, has two different computer automated feeding systems capable of capturing individual animal feeding behavior, the single animal Osborne FIRE feeder system and an in-house designed multi-animal feeder. Data collected from the FIRE feeder system, for nearly 9,000 animals, was collated and cleaned. Analyses to identify deviations in individual animal feeding behavior associated with health conditions is currently underway. We have begun to collect environmental data inside barns to improve our ability to incorporate the effects of air temperature and humidity in our predictive models as well as monitor environmental differences based on location in the barn. Another component of feed efficiency is fatty acid metabolism (Objective 1B). Fatty acid composition is a key factor in assessing the nutritional value of food and is a major determinant of pork quality and flavor. A genome-wide association study (GWAS) was performed on 818 pork samples for twenty-six fatty acid composition traits. Associations were found in or near seven genes (FADS2, ELOVL3, ELOVL5, ELOVL6, ELOVL7, SCD and HIF1AN) for sixteen fatty acid traits. Sequence variants in these genes and two other candidates were imputed and retrieved from the SWine IMputation (SWIM) Server. The combined gene regions spanning about 433 kilobases returned an additional 4,739 single nucleotide polymorphism (SNP). The combined GWAS increased the number of significant SNPs from 360 to 2,000 and from 16 to 17 traits in all 9 candidate genes. Nearly 40% of the most significant variants resided in functional regulatory sites, suggesting they modify gene expression. A new experiment to study the effects of social interactions among pen mates and chronic stress from these interactions on feeding behavior and feed utilization was initiated (Objective 1B). Well-being assessments, which include lesion scoring and overall assessment of the animal, are being conducted biweekly and at each mixing event from weaning up through the gilts’ entrance into the breeding herd. These data will also be used to look for associations of animal social status and maternal performance. We continue to study factors that will improve sow lifetime productivity (Objective 1C). The first step to having a productive sow is proper expression of estrus by gilts for successful breeding. Failure to express standing estrus occurs in about 10% of the gilts in most herds and approximately half of these gilts have had one or more ovulatory cycles without showing estrous behavior. These are referred to as behavioral anestrus gilts. We performed a case-control GWAS on 515 behavioral anestrus and 2,421 normally cycling control gilts identifying 81 significant SNP associations in 75 genomic regions. Sequence variants in twenty-one candidate genes were imputed and retrieved from SWIM. The combined gene regions spanned about 7.25 megabases and returned an additional 108,980 imputed SNP. The combined GWAS had significant SNP in nine candidate genes, seven of which have been associated with fertility traits in other species (RXRG, NTRK2, IGSF1, IGFS10, LRP1B, PLCB4 and NALCN). Once animals enter the breeding herd, many will be prematurely culled for becoming lame (Objective 1C). Therefore, a GWAS with fine mapping was conducted for electronic measures associated with structural soundness in gilts. First, a typical single-step best linear unbiased prediction analysis was conducted and then quantitative trait loci (QTL) regions were further interrogated with sequence-level SWIM genotypes. Several significant markers reside in genes involved in cartilage development, osteoarthritis progression or immune function. Estimates of genomic heritability ranged from 0.2-0.4 and genetic markers account for approximately 20% of the genomic heritability for most traits. As part of a collaborative project, we have utilized geophone sensors (vibration sensors) to monitor activity in farrowing pens. Measures that are related to sow productivity that have been tracked are initiation of farrowing, measurement of sow heart and respiration rate, and nursing behavior. With these novel phenotypes, we hope to be able to create a sow mothering ability index that can be used for selection of more attentive dams. This work addresses Objectives 1 and 2 of the project plan as developing the computer models to create these novel phenotypes utilize machine learning methodology.

Accomplishments
1. Discovery of genetic markers for soundness in pigs. Lameness in sows is a major reason for premature culling and the primary reason sows are euthanized in commercial swine production. If genetic traits of pigs associated with lameness could be discovered, then selection could be applied to reduce the incidence of lame sows, improving animal well-being. ARS scientists at Clay Center, Nebraska, and collaborators conducted a genomic analysis in a population of more than 3,000 pigs that had walking mobility measurements recorded by a pressure-sensing mat as well as 7 days of activity determined through computer-processed video recordings. Estimates of heritability were greatest for video recorded traits, but both sets of electronically captured traits were better than literature values published for data collected by trained observers. A genome-wide association analysis found several important regions of the genome controlling mobility traits and the most significant markers typically resided in genes associated with cartilage development, arthritis progression in humans or immune function. The genetic markers identified in this study will enable selection for improved structural soundness of breeding animals and provide insight into the progression of lameness in sows.

2. Health status of pigs predicted based on feeding behavior. Currently, swine producers rely solely on the skill level of animal caretakers to determine when pigs need management intervention. However, changes in the industry toward fewer and larger farms, coupled with a dwindling labor force, have made it difficult to recruit knowledgeable, dedicated animal caretakers. Precision livestock farming technologies, such as wearable sensors and cameras, are revolutionizing animal welfare by providing continuous animal monitoring, permitting detection of deviations from normality at the individual level, rather than understanding the “average” of animals in a pen or barn. For example, automated feeding systems have been developed to monitor feeding behavior of pigs. Disruptions in feeding behavior can be indicative of the onset of illness and/or other impairments of animal well-being. ARS researchers at Clay Center, Nebraska, developed a mathematical model for predicting daily time at the feeder for individual pigs in the grow-finish stage of production. The model was used to detect changes in feeding behavior associated with health status of the pig. Health issues were identified by the model, on average, 2.8 days earlier than diagnosis by the animal caretaker. Results from this work will serve as the basis for a decision support tool to assist producers in detecting illness and injury for individual pigs housed in large pens, potentially improving both swine production and animal wellbeing.

3. Shoulder lesions in sows detected with video processing. Shoulder lesions in lactating sows is a common issue in commercial swine production. Shoulder lesions, similar to bed sores in humans, reduce the animal’s well-being, can become infected and potentially result in premature culling of the sow. ARS researchers at Clay Center, Nebraska, and collaborators at the University of Nebraska, developed a video processing algorithm that can identify and measure shoulder lesions from video recorded with overhead cameras in a farrowing barn. This information was then used to identify associations with shoulder lesions and the production environment. Farrowing pen size as well as the use of one or two heat lamps were evaluated. It was determined that sows in the smallest farrowing pen (industry’s standard size) with two heat lamps had a much higher incidence of shoulder lesions than any other farrowing pen layout studied. In addition, older sows that were lighter weight for their age were also more prone to develop shoulder lesions. Detection of shoulder lesions in sows can be accomplished with computer processed video images to provide alerts to animal caretakers. In addition, this detection system would be extremely helpful to monitor and provide mitigations for at-risk sows (older, light-weight sows) or facilities with higher incidence of lesions due to its design. Implementation of these detection techniques will substantially reduce the incidence of shoulder lesions in adult sows.

4. Modified farrowing pen layout reduces piglet mortality. Preweaning mortality in commercial swine production ranges between 15-20% of all piglets born, representing a serious animal welfare issue and loss of production. While litter size has changed dramatically in the past 20 years, little research has evaluated changes in farrowing pen layouts. Modern sows are leaner, more productive and more prone to heat stress. ARS researchers at Clay Center, Nebraska, hypothesized offsetting the crate, rather than having it centered, in the farrowing pen would increase the distance between the heat lamp and sow, improving sow comfort and potentially piglet survival. Two novel layouts were tested, one where the crate was positioned diagonally in the farrowing pen and the other where the crate was positioned on a line one-third of the width of the pen. Both novel layouts tended to have fewer stillborn and crushed piglets in addition to better piglet performance. While the improvement was only about 1% lower mortality, it would lead to more than one million additional market pigs each year in the United States.

Review Publications
Bery, S., Brown-Brandl, T.M., Jones, B.T., Rohrer, G.A., Sharma, S.R. 2023. Determining the presence and size of shoulder lesions in sows using computer vision. Animals. 14(1). Article 131. https://doi.org/10.3390/ani14010131.
Brown-Brandl, T.M., Hayes, M.D., Rohrer, G.A., Eigenberg, R.A. 2022. Thermal comfort evaluation of three genetic lines of nursery pigs using thermal images. Biosystems Engineering. 225:1-12. https://doi.org/10.1016/j.biosystemseng.2022.11.002.
Ramirez, B.C., Hoff, S.J., Hayes, M.D., Brown-Brandl, T.M., Harmon, J.D., Rohrer, G.A. 2022. A review of swine heat production: 2003 to 2020. Frontiers in Animal Science. 3. Article 908434. https://doi.org/10.3389/fanim.2022.908434.
Dong, Y., Bonde, A., Codling, J.R., Bannis, A., Cao, J., Macon, A., Rohrer, G., Miles, J., Sharma, S., Brown-Brandl, T., Sangpetch, A., Sangpetch, O., Zhang, P., Noh, H. 2023. PigSense: Structural vibration-based activity and health monitoring system for pigs. ACM Transactions on Sensor Networks. 20(1):1-43. https://doi.org/10.1145/3604806.
Millburn, S., Schmidt, T., Rohrer, G.A., Mote, B. 2023. Identifying early-life behavior to predict mothering ability in swine utilizing NUtrack system. Animals. 13(18). Article 2897. https://doi.org/10.3390/ani13182897.
Keel, B.N., Lindholm-Perry, A.K., Rohrer, G.A., Oliver, W.T. 2023. Estimation of cell type proportions from bulk RNA-Seq of porcine whole blood samples using partial reference-free deconvolution. Animal Gene. 30. Article 200159. https://doi.org/10.1016/j.angen.2023.200159.
Wijesena, H.R., Nonneman, D.J., Rohrer, G.A., Lents, C.A. 2023. Relationships of genomic estimated breeding values for age at puberty, birth weight, and growth during development in normal cyclic and acyclic gilts. Journal of Animal Science. 101. Article skad258. https://doi.org/10.1093/jas/skad258.
Teng, J., Gao, Y., Yin, H., Bai, Z., Liu, S., Zeng, H., Consortium, T., Bai, L., Cai, Z., Zhao, B., Li, X., Xu, Z., Lin, Q., Pan, Z., Yang, W., Yu, X., Guan, D., Hou, Y., Keel-Mercer, B.N., Rohrer, G.A., Lindholm-Perry, A.K., Oliver, W.T., Ballester, M., Crespo, D., Quintanilla, R., Canela-Xandri, O., Rawlik, K., Xia, C., Yao, Y., Zhao, Q., Yao, W., Yang, L., Li, H., Zhang, H., Liao, W., Chen, T., Karlskov-Mortensen, P., Fredholm, M., Amills Eras, M., Clop, A., Giuffra, E., Wu, J., Cai, X., Diao, S., Pan, X., Wei, C., Li, J., Cheng, H., Wang, S., Su, G., Sahana, G., Lund, M., Dekkers, J., Kramer, L., Tuggle, C.K., Corbett, R., Groenen, M.A., Madsen, O., Godia, M., Rocha, D., Li, C., Pausch, H., Hu, X., Frantz, L., Luo, Y., Lin, L., Zhou, Z., Zhang, Z., Chen, Z., Cui, L., Xiang, R., Shen, X., Li, P., Huang, R., Tang, G., Li, M., Zhao, Y., Yi, G., Tang, Z., Jiang, J., Zhao, F., Yuan, X., Liu, X., Chen, Y., Xu, X., Zhao, S., Zhao, P., Haley, C., Zhou, H., Wang, Q., Pan, Y., Ding, X., Ma, L., Li, J., Navarro, P., Zhang, Q., Li, B., Tenesa, A., Liu, G. 2024. A compendium of genetic regulatory effects across pig tissues. Nature Genetics. 56:112-123. https://doi.org/10.1038/s41588-023-01585-7.
Wijesena, H.R., Keel, B.N., Nonneman, D.J., Cushman, R.A., Lents, C.A. 2023. Clustering of multi-tissue transcriptomes in gilts with normal cyclicity or delayed puberty reveals genes related to pubertal development. Biology of Reproduction. 110(2):261-274. https://doi.org/10.1093/biolre/ioad145.
Lindholm-Perry, A.K., Keel, B.N., Hales, K.E., Wells, J.E., Kuehn, L.A., Keele, J.W., Crouse, M.S., Nonneman, D.J., Nagaraja, T.G., Lawrence, T.E., Amachawadi, R.G., Carroll, J.A., Burdick Sanchez, N.C., Broadway, P.R. 2024. Ileal epithelial tissue transcript profiles of steers with experimentally induced liver abscesses. Applied Animal Science. 40(3):414-420. https://doi.org/10.15232/aas.2023-02503.
Cushman, R.A., Kaps, M., Snider, A.P., Crouse, M.S., Woodbury, B.L., Keel, B.N., McCarthy, K.L. 2024. Relationship of length of the estrous cycle to antral follicle number in crossbred beef heifers. Translational Animal Science. 8. Article txae074. https://doi.org/10.1093/tas/txae074.