A polygenic risk score (PRS), also known as a polygenic score or genetic risk score, is a number that quantifies the risk of an individual developing a particular trait or disease based on the cumulative effect of multiple genetic variants, typically thousands.1 Unlike Mendelian diseases, where a single mutation determines disease risk, most common diseases, such as heart disease, diabetes, and many psychiatric and orthopedic conditions, are influenced by many genes, each contributing a small effect.2 A prototypical polygenic trait that is easy to measure and has been widely studied in the development of this field is height.3,4 Educational attainment has also been a focus in humans.5
To gauge an individual's genetic risk for common diseases, researchers employ PRS values derived from genome-wide association study (GWAS) and individual genotype data within a separate target group. Polygenic risk score values provide a way to sum up an individual's risk alleles across the genome, with each allele's contribution weighted by its effect size.1 The foundational principles of PRS prediction are grounded in over a century of research in complex trait genetics and genetic risk prediction, with early implementation of predictive models notably seen in agriculture for calculating estimated breeding values in production animals.6 Just as the agricultural sector faced hurdles in applying genetic risk prediction across different breeds of domesticated animals, noting a drop in accuracy when applying them across genetically diverse ancestral populations,7 the application of PRS prediction across human populations with differing ancestry presented similar challenges.8 In theory, PRS values can be developed for any disease with a sufficiently high heritability. In human health, this research approach has been most widely described for cancer and coronary artery disease.9
Polygenic risk scores have been developed to predict a broad range of human diseases including cardiovascular diseases, various cancers, psychiatric disorders, neurological conditions, autoimmune diseases, and metabolic disorders.6 The application of the PRS approach is burgeoning across medical research, leveraging findings from GWAS to predict genetic risk in individuals for personalized health care. Recent findings that suggest the effectiveness and applicability of the PRS approach vary widely across different diseases and populations. This variability highlights that although identifying prospective PRS marker sets using GWAS association studies can be relatively straightforward, robust validation within and between breeds and populations remains challenging due to genetic diversity, population stratification, and breed-specific linkage disequilibrium structure.10–12 While PRS risk prediction shows promising predictive accuracy for several diseases with an important genetic contribution, prediction accuracy is currently highest for individuals of European ancestry due to the predominance of these populations in GWAS research, which provides very large data sets for model training. Ongoing research is aimed at enhancing the broader application of the PRS approach in more diverse populations for wider applicability and improved disease prediction across the global human population.13
In domesticated agricultural animals, the concept equivalent to PRS is more commonly referred to as estimated breeding values or genomic selection.14,15 Here PRS values predict an individual animal's genetic worth for a specific trait or set of traits important to farming, based on either pedigrees or both genotypic and pedigree data. In companion animals, phenotypic screening for common complex diseases has been used in orthopedics for many years, particularly screening for cruciate ligament rupture, hip dysplasia or elbow dysplasia in dogs, and degenerative suspensory ligament desmitis in horses.16–18 The use of PRS values for health screening in dogs, cats, and horses is still emerging but has shown promise for identifying genetic predispositions to certain diseases, such as hip dysplasia or cruciate ligament rupture in dogs, hypertrophic cardiomyopathy in cats, and degenerative suspensory ligament desmitis in horses. Polygenic risk score values can help breeders make informed decisions aimed at reducing the prevalence of genetic diseases in affected breeds.19 Moreover, they also offer insight for companion animal owners into the personalized health and care of their animals. Currently, the wider application of PRS prediction of disease risk in companion animals faces challenges such as diverse breed-specific genetic architectures in companion animal species and a lack of comprehensive large genetic and phenotypic databases. This review aims to comprehensively discuss progress in the development of PRS risk prediction for companion animal disease, the factors affecting generalized application, and promising areas for improving PRS prediction accuracy and implementation in companion animals.
Polygenic Risk Score Prediction in Companion Animals
where Xij is the jth column of the genotype matrix containing allelic counts of the risk allele (0, 1, or 2) for the ith individual, p is the total number of single nucleotide polymorphism (SNP) variants included in the score, and
The basic process of any PRS prediction starts with the creation of a training (reference) set population, ie, subjects who have both genotypic and phenotypic information. This information is then used to train a model, where the phenotype or disease status is used as a response and genotype and demographic data along with relevant nongenetic factors as predictors.3,21–24 Information from the predictive model is then used to estimate the risk liability of individuals in a test set of individuals having only genotypic and nongenetic information. The basic process of PRS prediction is also explained in Figure 1.
Polygenic risk score prediction (PRS) schematic in companion animals.
Citation: American Journal of Veterinary Research 2025; 10.2460/ajvr.25.01.0018
Candidate Traits for PRS Prediction in Companion Animals
As presented in Figure 2, various researchers have attempted to predict the risk of different complex diseases across various dog populations. In dogs, so far, PRS values have been generated for different hip dysplasia phenotypes in several studies.25–28 Hip dysplasia is a common orthopedic disease with moderate heritability estimates of approximately 0.37.29 In 2 different studies, different risk prediction models (pedigree best linear unbiased prediction [BLUP], genomic BLUP [GBLUP], single-step GBLUP, and Bayes C) were used that considered either pedigree or genome-wide SNP marker information or both to generate PRS values for the trait.25,29 Genome-based approaches, including GBLUP and Bayes C, provided similar higher accuracies than traditional pedigree-based methods, even with relatively small training sets. This suggests that PRS risk prediction is a promising strategy for improving breeding selection against hip dysplasia in dogs.25,29,30
Prediction accuracy of receiver operator characteristic scores for some important diseases in different dog breeds.
Citation: American Journal of Veterinary Research 2025; 10.2460/ajvr.25.01.0018
In 2017, a GWAS of cruciate ligament rupture in the Labrador Retriever was undertaken, and it was shown that PRS values in cruciate ligament rupture cases and controls were significantly different.16,31–33 Subsequently, the same research group expanded this analysis using 4 Bayesian and 4 machine learning models to generate PRS values for cruciate ligament rupture in Labrador Retrievers in the US. This research showed that both Bayesian and machine learning models could predict cruciate ligament rupture in the Labrador Retriever with a clinically relevant accuracy when covariates, including weight, sex, and neuter status, were included in the model. The best-performing model with a limited number of SNPs achieved an accuracy of 0.792. When a larger reference population is used, prediction is further improved, and use of multiple covariates is no longer needed.34
In another PRS prediction study35 of distichiasis in the Norwegian Staffordshire Bull Terrier, prediction accuracy was 0.66 using a GBLUP model. The investigators concluded that this level of prediction accuracy suggests a PRS prediction approach is suitable for the prediction of the risk of distichiasis in individual dogs of this breed. In another PRS risk prediction study36 of distichiasis in a different population of guide dogs that also used a GBLUP model, prediction accuracy was much higher at 0.94. The difference between the two studies35,36 is likely explained by differences in the genetic architecture between the 2 dog breeds, differences in the size or structure of the genetic data, or possibly better case-control matching. Additionally, PRS prediction accuracy was also assessed for acral lick dermatitis and mandibular distocclusion using 5 machine learning models in this second study.36 Prediction accuracy ranged from 0.59 to 0.69 and 0.68 to 0.81, respectively, for the 2 diseases.36 In another study,37 PRS prediction of osteosarcoma risk was undertaken in the Greyhound, Irish Wolfhound, and Rottweiler using the least absolute shrinkage and selection operator and stepwise regression models. The computed R2 values ranged from 0.377 to 0.741 using stepwise regression and were 0.24 for Irish Wolfhounds and 0.9 for Rottweilers using the least absolute shrinkage and selection operator regression model. Such breed differences can be attributed to the fact that dog populations are not geographically or temporally uniform, strong artificial selection within breeds, and breed-specific variation in genetic architecture, which includes differences in minor allele frequencies, genetic diversity, and the specific genetic risk factors associated with osteosarcoma.
These findings suggest the accuracy of PRS prediction models can vary substantially for different traits, likely due to differences in the genetic architecture of each disease. Choosing the appropriate PRS prediction model is crucial for maximizing prediction accuracy of different canine diseases. Collectively, these studies demonstrate that no single PRS prediction model is universally superior. Therefore, it is recommended that multiple models are considered, potentially in an ensemble to further improve prediction outcomes.
Multiancestry PRS Prediction in Dogs Belonging to Different Breeds
In humans, numerous studies have explored the use of PRS values as biomarkers in preventive medicine, yet considerable efforts are required to accurately determine and convey the absolute risk associated with intrinsic genetic and extrinsic (lifestyle) risk factors to patients from differing ancestry.20,38,39 Most of the significant challenges facing PRS risk prediction today lie in limited applicability for accurate prediction in multiancestry (MA) populations. In companion animals, preliminary research in dogs suggests the same problem exists (Figure 3). To address this problem in companion animals, more research is needed to enhance PRS generalizability, including development of new bioinformatics approaches in domesticated species with a breed structure.
Breed-specific prevalence of cruciate ligament rupture, osteosarcoma, distichiasis, and hip dysplasia in dogs.
Citation: American Journal of Veterinary Research 2025; 10.2460/ajvr.25.01.0018
In human genetics, PRS values developed from MA GWAS in populations with differing ethnicities have shown promising potential for improving PRS prediction accuracy and generalizability across populations. One reason could be related to the substantial contribution of shared causal variants or even mosaic ancestries to the heritable variation of complex traits and diseases that are often shared across ancestral populations.8,40,41
To achieve an accurate genomic prediction in livestock or production animals, such as dairy cattle, one strategy involves leveraging SNP effects from different breeds, in an across-breed prediction scenario, when a large reference population is available, or incorporating data from various breeds in a multibreed prediction scenario to enhance the precision of SNP effect estimates. Combining data from several breeds into a MA reference population often considers homogeneous SNP effects across all breeds.42,43 However, this relies on SNP-quantitative trait loci (QTL) linkage disequilibrium (LD) across breeds, which might be different, or the phase of the SNP and QTL alleles may be reversed among the breeds, due to selection and genetic drift. However, different breeds and their crosses do not necessarily have the same QTL effects. Furthermore, even if the QTL effects are the same, the LD between SNP markers and disease-associated QTL may differ.31,42,44–46 Therefore, a MA prediction model that does not account for differences in SNP effects between breeds may not be accurate, especially if the breeds are relatively distant from each other.
Polygenic risk score methodologies and factors such as heterogeneity of pathogenetic mechanisms, breed structure, including comparisons between different within-breed populations, LD, minor allele frequencies, heritability, sample sizes, genetic correlation, and marker differences between reference and test populations all influence prediction accuracy, and the optimal approach has remained largely unclear in PRS prediction in companion animals. In 1 of our studies,34 we compared GWAS results from Labrador Retriever populations with those from Rottweiler populations, varying sample sizes and anterior cruciate ligament rupture prevalence (5.8% vs 7.8%) to simulate MA GWASs with different ancestry compositions. Using a 10-fold cross-validation model, we showed that a MA training set can enhance prediction accuracy for highly polygenic traits.34
Phenotyping Medical Records and Biobank Repositories for Companion Animals
Phenotypic data recording offers numerous advantages for veterinary care and research, including a better understanding of disease prevalence, broader sampling, and temporal analysis of health trends. When combined with genetic data, it enhances development of PRS scores, helping predict disease risk and inform breeding and treatment decisions.
During recent years, the veterinary informatics sector has struggled with the curation of large-scale datasets like Vet Compass, managed by the Royal Veterinary College in the UK, and the Small Animal Veterinary Surveillance Network (SAVSNET), which offer a wealth of canine health records but face challenges due to limited coding options and variable data quality from clinicians and the Canine Health Information Center, which is managed by the Orthopedic Foundation for Animals. Therefore, there is a clear opportunity to leverage these databases in genomic research.47,48 Extrapolating from human studies shows great promise for the application of electronic medical recording in modern human healthcare, but these technologies must be directly applied to companion animals to understand their true value in advancing companion animal healthcare.
In addition, the Dog10K Genome Project represents a large leap forward in canine genomics, aiming to sequence and analyze over 10,000 canid genomes. This international effort has already generated 20X coverage sequencing data from 1,987 individuals, including a wide range of dog breeds, village dogs, wolves, and coyotes. The dataset reveals over 48 million variants, offering the most extensive catalog of canine genetic variation to date.49 The project provides detailed insights into breed-specific genetic architecture. Over time, this database should enable more precise PRS prediction and improve understanding of breed-specific disease susceptibility. So far, the Dog10K project has identified a vast array of SNPs, indels, and structural variants across the canid family. This comprehensive catalog is crucial for improving the accuracy of PRS and enhancing GWAS by providing a more complete picture of genetic variation. The expectation is that the integration of phenotypic and genetic information from comprehensive canine biobanks like Vet Compass, Small Animal Veterinary Surveillance Network, Canine Health Information Center, and the Dog10K Genome Project will lead to a significant promise for advancing veterinary care and enhance PRS accuracy and deepen our understanding of disease susceptibility prediction in companion animals.
Computational Models for Predicting Risk of Complex Diseases in Companion Animals
Accurately predicting complex diseases in companion animals is crucial for effective personalized treatment as in human medicine. Polygenic risk score prediction requires the development of statistical methods that can properly model the complex polygenic architecture of the diseases using high-throughput genome-wide variants.10,50 The aim of an accurate prediction model is to capture the most genetic variance for a given disease. To maximize the captured variance and improve transferability of the estimated PRS, statistical and computational methods innovation is needed to (1) model genetic markers across the entire genome simultaneously, ensuring efficient use of all available data while considering local LD patterns; (2) account for varying effect sizes across a range of complex traits and diseases, from highly polygenic conditions (eg, height, schizophrenia) to those with a combination of small effects and clusters of loci with moderate to large impacts (eg, autoimmune disorders, Alzheimer disease); (3) maintain computational efficiency and scalability; (4) integrate environmental and lifestyle factors to account for gene-environment interactions, allowing for more personalized predictions of disease risk based on both genetic predisposition and external influences; and (5) incorporate multiomics data, such as transcriptomics, epigenomics, and proteomics, to capture a more comprehensive biological picture, which can enhance the predictive accuracy by linking genetic variants to functional outcomes and downstream biological processes.11
Due to the high diversity in companion animal populations like dog breeds, it is critical to consider the transferability of PRS values across different populations because the genetic architectures can vary significantly across populations, making it necessary to evaluate how well PRS values constructed from one breed perform in another. Without considering transferability, a PRS prediction approach developed for one population might not perform well when applied to individuals from a different genetic background due to differences in LD patterns, allele frequencies, and trait genetic architectures.41,51–53 If PRS values are to be effectively used for disease screening, personalized treatment, and precision medicine across global populations, they must be evaluated for their robustness across diverse ancestries. This consideration is key to avoiding health disparities in genetic risk predictions and ensuring that all populations benefit equally from advances in genetic prediction models.
Limitation of Potential Health Benefits of PRS Prediction in Companion Animals
To date, use of PRS values to predict disease risk in companion animals has seen limited implementation when compared to the use of this approach in livestock or human health. Several factors contribute to this limited application as described in the following sections.
Genetic diversity
Companion animals, especially dogs, have a high level of genetic diversity, particularly between breeds.54–56 Such diversity can complicate the development of accurate PRS prediction models that are broadly applicable for risk prediction in individual dogs, particularly if the dog has an ancestry or breed background that is different from the reference population used for training the predictive algorithm. For example, the findings from a recent study57 indicated that a specific genetic marker, a 16-bp deletion in the pyruvate dehydrogenase kinase 4 gene, which was previously associated with dilated cardiomyopathy in a US cohort of Doberman Pinschers, did not show the same association in a European cohort. This discrepancy highlights the importance of validating genetic markers across different populations before they can be deemed reliable for broader application.57
Research investment
There has been significantly more investment in genomic research for livestock and human health due to the direct economic benefits to agriculture and the potential for improving human health outcomes. In contrast, companion animal genomics has attracted less funding, limiting the scope of research and the development of comprehensive genomic databases.
Breed-specific disease
Many companion animals, particularly dogs, are prone to breed-specific diseases. While this aspect of canine health could theoretically make it easier to investigate the genetic contribution to disease risk, it also means that PRS prediction accuracy developed for one breed may not be applicable to another, requiring breed-specific studies58 that further increase the cost and complexity of research. For example, a recent study58 using PRS prediction in Korean Sapsaree dogs found 194 significant SNPs associated with morphological traits and hip laxity measured by the distraction index, a critical risk factor for hip dysplasia. Estimated predictive accuracies were specific to the Sapsaree breed and may not have the sample predictive power in other breeds, emphasizing the need for breed-specific studies.58
Availability of data
The successful development of PRS prediction relies on access to large genomic datasets and detailed health records. For companion animals, such datasets are less common and often less detailed than those available for humans or economically significant livestock. This lack of high-quality large biobank data hampers the ability to identify genetic markers associated with diseases or traits across diverse animal populations.
Ethical and regulatory considerations
There are ethical considerations regarding the use of genetic selection in companion animals, especially concerning the potential to exacerbate existing health issues related to extreme breed traits. In addition, the regulatory landscape for genetic testing in pets is less clear, potentially hindering the development and commercialization of PRS risk-based tests.
Market demand and awareness
Awareness and demand for genetic risk testing among companion animal owners and breeders are growing but remain relatively limited compared to the interest in human genetic testing. The perceived value of PRS prediction of disease risk in improving the health and well-being of companion animals is increasing, but widespread acceptance and understanding are still developing.
In conclusion, despite several challenges with PRS prediction of disease risk in companion animals, interest in applying genomic prediction and PRS in companion animal healthcare is growing. Advances in genomics technology, increasing collaboration between veterinarians and genetic researchers, and a growing repository of genetic data are beginning to overcome some of the barriers to the broader application of PRS prediction of disease risk in companion animals. This progress holds promise for improving the health, longevity, and quality of life of companion animals through more personalized and preventative care strategies.
Acknowledgments
None reported.
Disclosures
The authors have nothing to disclose. No AI-assisted technologies were used in the composition of this manuscript.
Funding
The authors have nothing to disclose.
References
- 1.↑
Hingorani AD, Gratton J, Finan C, et al. Performance of polygenic risk scores in screening, prediction, and risk stratification: secondary analysis of data in the polygenic score catalog. BMJ Med. 2023;2(1):e000554. doi:10.1136/bmjmed-2023-000554
- 2.↑
Schork NJ. Genetics of complex disease: approaches, problems and solutions. Am J Respir Crit Care Med. 1997;156(4):S103. doi:10.1164/ajrccm.156.4.12-tac-5-S109
- 3.↑
Lello L, Avery SG, Tellier L, Vazquez AI, de Los Campos G, Hsu SDH. Accurate genomic prediction of human height. Genetics. 2018;210(2):477–497. doi:10.1534/genetics.118.301267-497
- 4.↑
Wood AR, Esko T, Yang J, et al. Defining the role of common variation in the genomic and biological architecture of adult human height. Nat Genet. 2014;46(11):1173. doi:10.1038/ng.3097-1186
- 5.↑
Lee JJ, Wedow R, Okbay A, et al. Gene discovery and polygenic prediction from a 1.1-million-person GWAS of educational attainment. Nat Genet. 2018;50(8):1112. doi:10.1038/s41588-018-0147-3
- 6.↑
Wray NR, Kemper KE, Hayes BJ, Goddard ME, Visscher PM. Complex trait prediction from genome data: contrasting EBV in livestock to PRS in humans. Genetics. 2019;211(4):1131. doi:10.1534/genetics.119.301859
- 7.↑
Burrow HM, Mrode R, Mwai AO, Coffer MP, Hayes BJ. Challenges and opportunities in applying genomic selection to ruminants owned by smallholder farmers. Agriculture. 2021;11(11):1172. doi:10.3390/agriculture11111172
- 8.↑
Wang Y, Tsuo K, Kanai M, Neale BM, Martin AR. Challenges and opportunities for developing more generalizable polygenic risk scores. Annu Rev Biomed Data Sci. 2022;5:293–320. doi:10.1146/annurev-biodatasci-111721-074830
- 9.↑
Sud A, Horton RH, Hingorani AD, et al. Realistic expectations are key to realising the benefits of polygenic scores. BMJ. 2023;380:e073149. doi:10.1136/bmj-2022-073149
- 10.↑
Torkamani A, Wineinger NE, Topol EJ. The personal and clinical utility of polygenic risk scores. Nat Rev Genet. 2018;19(9):581–590. doi:10.1038/s41576-018-0018-x
- 11.↑
Kachuri L, Chatterjee N, Hirbo J, et al. Principles and methods for transferring polygenic risk scores across global populations. Nat Rev Genet. 2024;25(1):8–25. doi:10.1038/s41576-023-00637-2
- 12.↑
Hou K, Xu Z, Ding Y, et al. Calibrated prediction intervals for polygenic scores across diverse contexts. Nat Genet. 2024;56(7):1386–1396. doi:10.1038/s41588-024-01792-w
- 13.↑
Weissbrod O, Kanai M, Shi H, et al. Leveraging fine-mapping multipopulation training data to improve cross-population polygenic risk scores. Nat Genet. 2022;54(4):450–458. doi:10.1038/s41588-022-01036-9
- 14.↑
Meuwissen TH, Hayes BJ, Goddard ME. Prediction of total genetic value using genome-wide dense marker maps. Genetics. 2001;157(4):1819–1829. doi:10.1093/genetics/157.4.1819
- 15.↑
Momen M, Mehrgardi AA, Sheikhi A, et al. Predictive ability of genome-assisted statistical models under various forms of gene action. Sci Rep. 2018;8(1):12309. doi:10.1038/s41598-018-30089-2
- 16.↑
Baker LA, Momen M, Chan K, et al. Bayesian and machine learning models for genomic prediction of anterior cruciate ligament rupture in the canine model. G3 (Bethesda). 2020;10(8):2619–2628. doi:10.1534/g3.120.401244
- 17.
Binversie EE, Momen M, Rosa GJM, Davis BW, Muir P. Across-breed genetic investigation of canine hip dysplasia, elbow dysplasia, and anterior cruciate ligament rupture using whole-genome sequencing. Front Genet. 2022;13:913354. doi:10.3389/fgene.2022.913354
- 18.↑
Momen M, Brounts SH, Binversie EE, et al. Selection signature analyses and genome-wide association reveal genomic hotspot regions that reflect differences between breeds of horse with contrasting risk of degenerative suspensory ligament desmitis. G3 (Bethesda). 2022;12(10):jkac179. doi:10.1093/g3journal/jkac179
- 19.↑
Momen M, Brauer K, Patterson MM, et al. Genetic architecture and polygenic risk score prediction of degenerative suspensory ligament desmitis (DSLD) in the Peruvian Horse. Front Genet. 2023;14:1201628. doi:10.3389/fgene.2023.1201628
- 20.↑
Polygenic Risk Score Task Force of the International Common Disease Alliance. Responsible use of polygenic risk scores in the clinic: potential benefits, risks and gaps. Nat Med. 2021;27(11):1876–1884. doi:10.1038/s41591-021-01549-6
- 21.↑
Jensen J, Su G, Madsen P. Partitioning additive genetic variance into genomic and remaining polygenic components for complex traits in dairy cattle. BMC Genet. 2012;13:44. doi:10.1186/1471-2156-13-44
- 22.
Do CB, Hinds DA, Francke U, Eriksson N. Comparison of family history and SNPs for predicting risk of complex disease. PLoS Genet. 2012;8(10):e1002973. doi:10.1371/journal.pgen.1002973
- 23.
Chatterjee N, Shi J, García-Closas M. Developing and evaluating polygenic risk prediction models for stratified disease prevention. Nat Rev Genet. 2016;17(7):392–406. doi:10.1038/nrg.2016.27
- 24.↑
Bulik-Sullivan BK, Loh P-R, Finucane HK, et al. LD score regression distinguishes confounding from polygenicity in genome-wide association studies. Nat Genet. 2015;47(3):291–295. doi:10.1038/ng.3211
- 25.↑
Guo G, Zhou Z, Wang Y, et al. Canine hip dysplasia is predictable by genotyping. Osteoarthritis Cartilage. 2011;19(4):420–429. doi:10.1016/j.joca.2010.12.011
- 26.
Lavrijsen ICM, Heuven HCM, Meij BP, et al. Prevalence and co-occurrence of hip dysplasia and elbow dysplasia in Dutch pure-bred dogs. Prev Vet Med. 2014;114(2):114–122. doi:10.1016/j.prevetmed.2014.02.001
- 27.
Mikkola L, Holopainen S, Pessa-Morikawa T, et al. Genetic dissection of canine hip dysplasia phenotypes and osteoarthritis reveals three novel loci. BMC Genomics. 2019;20(1):1027. doi:10.1186/s12864-019-6422-6
- 28.↑
Lewis TW, Blott SC, Woolliams JA. Genetic evaluation of hip score in UK labrador retrievers. PLoS One. 2010;5(10):e12797. doi:10.1371/journal.pone.0012797
- 29.↑
Edwards SM, Woolliams JA, Hickey JM, et al. Joint genomic prediction of canine hip dysplasia in UK and US Labrador Retrievers. Front Genet. 2018;9:101. doi:10.3389/fgene.2018.00101
- 30.↑
Jiang L, Li Z, Hayward JJ, et al. Genomic prediction of two complex orthopedic traits across multiple pure and mixed breed dogs. Front Genet. 2021;12:666740. doi:10.3389/fgene.2021.666740
- 31.↑
Lee B, Baker L, Momen M, et al. Identification of genetic variants associated with anterior cruciate ligament rupture and AKC standard coat color in the Labrador Retriever. BMC Genom Data. 2023;24:60. doi:10.1186/s12863-023-01164-z
- 32.
Baker L, Momen M, Mcnally R, et al. Biologically enhanced genome-wide association study provides further evidence for candidate loci and discovers novel loci that influence risk of anterior cruciate ligament rupture in the dog model. Front Genet. 2021;12:200. doi:10.3389/fgene.2021.593515
- 33.↑
Binversie EE, Baker LA, Engelman CD, et al. Analysis of copy number variation in dogs implicates genomic structural variation in the development of anterior cruciate ligament rupture. PLoS One. 2020;15(12):e0244075. doi:10.1371/journal.pone.0244075
- 34.↑
Momen M, Kearney HK, Patterson MM, et al. Cross-species analysis of genetic architecture and polygenic risk scores for non-contact ACL rupture in dogs and humans. Commun Biol. 2025;8(1):26. doi:10.1038/s42003-024-07395-9
- 35.↑
Jørgensen D, Ropstad E-O, Meuwissen T, Lingaas F. Genomic analysis and prediction of genomic values for distichiasis in Staffordshire bull terriers. Canine Med Genet. 2023;10(1):9. doi:10.1186/s40575-023-00132-1
- 36.↑
Thorsrud JA, Evans KM, Holle DM, et al. Comparison of GBLUP and machine learning breeding values for binary disease traits in working dogs. In: Proceedings of 12th World Congress on Genetics Applied to Livestock Production (WCGALP) Technical and Species Orientated Innovations in Animal Breeding, and Contribution of Genetics to Solving Societal Challenges. Wageningen Academic Publishers; 2022:1643–1646.
- 37.↑
Zapata I, Moraes LE, Fiala EM, et al. Risk-modeling of dog osteosarcoma genome scans shows individuals with Mendelian-level polygenic risk are common. BMC Genomics. 2019;20(1):226. doi:10.1186/s12864-019-5531-6
- 38.↑
Igo RP, Kinzy TG, Cooke Bailey JN. Genetic risk scores. Curr Protoc Hum Genet. 2019;104(1):e95. doi:10.1002/cphg.95
- 39.↑
Crouch DJM, Bodmer WF. Polygenic inheritance, GWAS, polygenic risk scores, and the search for functional variants. Proc Natl Acad Sci U S A. 2020;117(32):18924–18933. doi:10.1073/pnas.2005634117
- 40.↑
Márquez-Luna C, Loh P, South Asian Type 2 Diabetes (SAT2D) Consortium; SIGMA Type 2 Diabetes Consortium; Price AL. Multiethnic polygenic risk scores improve risk prediction in diverse populations. Genet Epidemiol. 2017;41(8):811–823. doi:10.1002/gepi.22083
- 41.↑
Cavazos TB, Witte JS. Inclusion of variants discovered from diverse populations improves polygenic risk score transferability. HGG Adv. 2021;2(1):100017. doi:10.1016/j.xhgg.2020.100017
- 42.↑
Hayes BJ, Bowman PJ, Chamberlain AJ, Goddard ME. Invited review: genomic selection in dairy cattle: progress and challenges. J Dairy Sci. 2009;92(2):433–443. doi:10.3168/jds.2008-1646
- 43.↑
Clasen JB, Fikse WF, Su G, Karaman E. Multibreed genomic prediction using summary statistics and a breed-origin-of-alleles approach. Heredity (Edinb). 2023;131(1):33–42. doi:10.1038/s41437-023-00619-4
- 44.↑
Vandenplas J, Calus MPL, Sevillano CA, Windig JJ, Bastiaansen JW. Assigning breed origin to alleles in crossbred animals. Genet Sel Evol. 2016;48(1):61. doi:10.1186/s12711-016-0240-y
- 45.
Begum F, Ghosh D, Tseng GC, Feingold E. Comprehensive literature review and statistical considerations for GWAS meta-analysis. Nucleic Acids Res. 2012;40(9):3777–3784. doi:10.1093/nar/gkr1255
- 46.↑
Gebreyesus G, Buitenhuis AJ, Poulsen NA, et al. Combining multi-population datasets for joint genome-wide association and meta-analyses: the case of bovine milk fat composition traits. J Dairy Sci. 2019;102(12):11124–11141. doi:10.3168/jds.2019-16676
- 47.↑
Neethirajan S, Kemp B. Digital phenotyping in livestock farming. Animals. 2021;11(7):2009. doi:10.3390/ani11072009
- 48.↑
Noble PJ, Farrell SO, Al-Moubayed N, Radford AD. Comprehensive representation of health-related phenotypes in one million dogs using topic modelling of electronic health records. ResearchSquare. Preprint posted online July 15, 2024. doi:10.21203/rs.3.rs-4621991/v1
- 49.↑
Meadows JRS, Kidd JM, Wang G-D, et al. Genome sequencing of 2000 canids by the Dog10K consortium advances the understanding of demography, genome function and architecture. Genome Biol. 2023;24(1):187. doi:10.1186/s13059-023-03023-7
- 50.↑
Ma Y, Zhou X. Genetic prediction of complex traits with polygenic scores: a statistical review. Trends Genet. 2021;37(11):995–1011. doi:10.1016/j.tig.2021.06.004
- 51.↑
Fohler L, Mayr A, Maj C, et al. Transferability of polygenic risk scores depending on demography and dominance coefficients. medRxiv. Preprint posted online August 8, 2024. doi:10.1101/2024.08.07.24311645.
- 52.
Kamiza AB, Toure SM, Vujkovic M, et al. Transferability of genetic risk scores in African populations. Nat Med. 2022;28(6):1163–1166. doi:10.1038/s41591-022-01835-x
- 53.↑
Zhao Z, Fritsche LG, Smith JA, Mukherjee B, Lee S. The construction of cross-population polygenic risk scores using transfer learning. Am J Hum Genet. 2022;109(11):1998–2008. doi:10.1016/j.ajhg.2022.09.010
- 54.↑
Bergström A, Stanton DWG, Taron UH, et al. Grey wolf genomic history reveals a dual ancestry of dogs. Nature. 2022;607(7918):313–320. doi:10.1038/s41586-022-04824-9
- 55.
Ostrander EA, Wayne RK, Freedman AH, Davis BW. Demographic history, selection and functional diversity of the canine genome. Nat Rev Genet. 2017;18(12):705–720. doi:10.1038/nrg.2017.67
- 56.↑
Anderson RS, Chu AK, Rylander H, et al. Pathologic classification of a late-onset peripheral neuropathy in a spontaneous Labrador retriever dog model. J Comp Neurol. 2024;532(3):e25596. doi:10.1002/cne.25596
- 57.↑
Owczarek-Lipska M, Mausberg T, Stephenson H, Dukes-McEwan J, Wess G, Leeb T. A 16-bp deletion in the canine PDK4 gene is not associated with dilated cardiomyopathy in a European cohort of Doberman Pinschers. Anim Genet. 2013;44(2):239–239. doi:10.1111/j.1365-2052.2012.02396.x
- 58.↑
Haque MA, Kim N-K, Yeji R, et al. Genomic prediction and genome-wide association studies of morphological traits and distraction index in Korean Sapsaree dogs. PLoS One. 2024;19(11):e0312583. doi:10.1371/journal.pone.0312583
