Skip to main content
NCBI home page
Wiley Open Access Collection logoLink to Wiley Open Access Collection
. 2026 Jun 17;57(3):e70146. doi: 10.1002/age.70146

Polygenic Prediction of Equestrian Sport Discipline Among Horses Bred for Jumping and Dressage

Emmeline W Hill 1,2,✉, Haige Han 1, Amy R Holtby 1, Heinrich Anhold 3, Beatrice A McGivney 1
PMCID: PMC13276299  PMID: 42310950

ABSTRACT

Horses bred for different equestrian sports vary in physical, physiological and behavioural requirements. Characterising genetic markers associated with discipline provides an opportunity to improve identification of horses best suited to either jumping or dressage. Using a composite selection signals test, genomic regions under positive selection in German warmblood horses divergent for discipline were identified that contained lead SNPs proximal to or within genes with functions relevant to the requirements for jumping and dressage. Dressage genes function in tortuosity of movement (GRM5) and responsiveness to touch (PROKR2), and have roles in bone development (VWC2, THSD7B, CYP27A1) and growth traits (VPS13B, TMEFF2, NANOS1, ZEB2). Jumping genes function in the modulation of neuropathic pain (LPAR6 and SLITRK1) and encode compounds commonly included in nutritional supplements (CHST15 and TPH2). A polygenic prediction score (PPS) derived from 28 SNPs had an AUC = 0.89, and categorical predictions for dressage and jumping had an overall accuracy of 80%. Among horses with jumping competition records, the jumping genetic profile was highest among elite performers (85.7%) and lowest among inferior performers (69.4%). The PPS could be applied to efficiently and accurately identify young horses most suitable for each discipline before they reach the age for participation in physical performance assessments, and the genetic markers could be applied in genome‐enabled breeding strategies to accelerate genetic gain in sport horses. Furthermore, examination of the PPS among Thoroughbred horses indicated that it could be used to inform suitable second careers for racehorses to improve welfare outcomes following retirement from racing.

Keywords: dressage, genetic association, genomics, horse, jumping, polygenic prediction, sport

1. Background

Many different breeds of horse are bred for equestrian sports with European warmblood horses registered to a range of different stud books excelling in certain disciplines. Different stud books lead in the World Breeding Federation for Sport Horses Studbook Rankings for jumping (i.e., Selle Francais, Zangersheide, Belgian Warmblood, Holsteiner) dressage (i.e., Dutch Warmblood, Oldenburger, Westphalian, Hanoverian) and eventing (i.e., Holsteiner, Irish Sport Horse, Selle Francais, Dutch Warmblood). Stud books are maintained for the registration and genetic improvement of sport horses, and although European warmblood horses are registered in different stud books, there is little genetic distinction among them, with most conforming to a generalised pan‐European warmblood genotype (Lindsay‐McGee et al. 2023; McGivney et al. 2023). This suggests that rather than maintaining distinct lineages, breeding for sporting discipline is the target for European sport horse breeders.

In genetic improvement programmes, stud books employ linear profiling for the breeding and selection of horses based on movement, conformation and jumping profiles, with different standards evaluated for dressage and jumping (Borowska and Lewczuk 2023). Furthermore, many stud books now estimate (genomic) breeding values (Doyle et al. 2022), predicated on the recognition that athletic potential has a heritable component. For jumping, heritability estimates vary depending on the traits measured, and as a result, heritability varies greatly for different stud books. The highest heritabilities for jumping and dressage have been estimated as 0.28–0.38 and 0.21–0.24, respectively (Welker et al. 2018).

Several genome‐wide association studies (GWAS) have been reported for jumping traits in warmbloods originating from different stud books (Schroder et al. 2012; Brard and Ricard 2015; Nazari‐Ghadikolaei et al. 2025). These studies have identified quantitative trait loci (QTLs) encompassing genes with functions in body height/stature, cardiac health, skeletal muscle development and metabolism, and depth perception.

Population genomics‐based selection signals approaches have been highly successful in identifying major genes for phenotypic traits in livestock (Randhawa et al. 2014, 2015), and the identification of genomic selection signals in warmblood horses has led to an understanding of genes that are both unique to and shared across breeds (Ablondi et al. 2019; Nolte et al. 2019). Within stud books, specialisation for different sporting disciplines has led to genetic variation that contributes to discipline type (Rovere et al. 2015, 2017; Heuer et al. 2016; Ablondi et al. 2019; Bonow et al. 2023).

Here, we used a population genomics‐based approach to identify genomic regions under selection for jumping and dressage disciplines in German warmblood horses. As positive selection increases the frequency of advantageous gene variants (Hoekstra et al. 2001; Grossman et al. 2010), we hypothesised that selection signals contain genetic variants associated with sporting discipline that could be combined in a polygenic risk (prediction) score (PPS) (Choi et al. 2020) to predict sporting discipline among sport horses.

2. Materials and Methods

2.1. Animal Cohorts

Details for the study cohorts are provided in Table S1.

2.1.1. Discovery Set

Horses were German warmblood males born in 2020–2022 that were bred and/or raised at the Brandenburg State Stud (Germany) with the aim to present them for stallion licensing at an inspection event. At the time of the study none was breeding. Horses were partitioned into two comparator cohorts, Dressage (n = 17) and Jump (n = 21) (Table S1), classified based on the pedigree and performance of their sire and dam, and on physical type (assessed in 2024). Dressage horses were the progeny of 16 individual sires and 16 individual dams, and Jump horses were the progeny of 13 individual sires and 19 individual dams.

2.1.2. Training Set

Horses were German warmblood males born in 2023 and females born in 2022–2023 that were bred and/or raised at the Brandenburg State Stud (Germany) and comprised n = 26 Dressage (n = 14 female, n = 12 male) and n = 29 Jump (n = 15 female, n = 14 male) classified based on the pedigree and performance of their sire and dam and on physical type (assessed in 2025). There was < 9% overlap of shared sires between the Discovery and Training sets.

2.1.3. Validation Sets

Horses with jumping competition records were defined as elite (5‐star performers, n = 14 European warmbloods), superior (2 & 3‐star performers, n = 15 Irish Sport Horses), and inferior (failed to achieve double clear at 1 M, n = 49 Irish Sport Horses). Genotypes for Thoroughbred horses originating from Australia (n = 1538), Europe (n = 625) and USA (n = 1270) were accessed from a commercial genetic archive. Genotypes for n = 14 Irish Sport Horses registered with Eventing Ireland were accessed from a commercial genetic archive.

2.2. Genotyping and QC

DNA was extracted from either blood or mane/tail hair samples using the Qiagen DNeasy Blood & Tissue Kit. Genotyping was performed using the Equine Illumina GGP genotyping array (Neogen). Quality control was performed to include samples and SNPs with a call rate > 99% and SNPs with a minor allele frequency > 5%. 55 559 SNPs were retained for analyses.

2.3. Principal Component Analysis

Principal component analysis (PCA) was performed using LD pruned SNPs (‐‐indep‐pairwise 100 5 0.2) and the –pca 10 function in PLINK 1.9 (Chang et al. 2015). Principal components (PCs) were plotted with points colour coded based on discipline (Dressage/Jump) and research set (Discovery/Training).

2.4. Composite Selection Signals Analyses

Composite selection signals (CSS) analyses (Randhawa et al. 2014, 2015) were performed using Jump and Dressage horses (Discovery set) separately as the target (selected) cohorts. CSS uses fractional ranks of constituent tests allowing a combination of the evidence of selection from different population genetic tests. CSS uses the fixation index (F ST), the change in selected allele frequency (ΔSAF) and the cross‐population extended haplotype homozygosity (XP‐EHH) tests and combines each test statistic into one composite CSS statistic for each SNP. For each constituent method, test statistics were ranked (1, …, n) genome‐wide on n SNPs. Ranks were converted to fractional ranks (r′) (between 0 and 1) by 1/(n + 1) through n/(n + 1). Fractional ranks were converted to z‐values as z = Φ−1(r′), where Φ−1(·) is the inverse normal cumulative distribution function. Mean z scores were calculated by averaging z‐values across all constituent tests at each SNP position and p‐values were directly obtained from the distribution of means from a normal N (0, m−1) distribution where m is the number of constituent test statistics. Log‐transformed p values (−log10 of p values of the mean z‐values) were declared as CSS. SNPs with extreme test scores (top 1%) in the genome‐wide distribution were considered significant. To minimize the spurious noise from single SNPs and tests with resultant false positives, the individual test statistics were then averaged across SNPs within 1 MB sliding windows (smoothed CSS score). Regions of interest (ROIs) were defined as clusters of ≥ 3 SNPs among the top 1% SNPs.

2.5. Polygenic Prediction Score (PPS) Test Development

SNPs in the CSS ROIs were extracted, and tests of genetic association were performed in PLINK 1.9 (Chang et al. 2015). The top SNP (lowest p‐value) within each cluster was identified for each comparison (Jump vs. Dressage and Dressage vs. Jump) and defined as the lead SNP for each ROI. The p‐value for significance was calculated based on a Bonferroni correction for multiple tests.

Using the odds ratio (OR) from the association analysis to assign SNP weightings, the profile function in PLINK was used to generate a polygenic risk score (PRS) model in the Training set for five SNP sets based on p‐value cut‐offs < 0.05, < 0.01, < 0.001, < 0.0001 and < 0.00001. The area under the curve (AUC) and two‐tailed t‐tests were used to determine the optimal model. As the phenotype is not a disease state, the PRS is referred to as a polygenic prediction score (PPS). To avoid overinflation of the PPS, two SNPs were excluded because the OR was calculated as 0 (i.e., the minor allele was not observed in one of the cohorts). The obtained scores were scaled to values ranging from 1 to 10 (the score/maximum score × 10). The median PPS in the Training set was used as the cut‐off value to predict Dressage (< median) or Jump (> median), and category ranges were further assigned as follows: D Pro 0–2.49; D 2.5–4.9; J 5–7.49; J Pro 7.5–10.

2.6. Validation of the PPS

To validate the PPS, the PPS was calculated for n = 78 horses with jumping competition records, and each individual was assigned to a prediction category. Horses with dressage competition records were not available for validation of the PPS for competition level in dressage. The PPS and prediction categories were also determined for n = 3433 Thoroughbred horses and n = 14 event horses.

3. Results

3.1. Population Genetic Structure

The population genetic structure of the Discovery and Training sets was examined following partitioning individuals into Jump or Dressage cohorts. In the PCA plot, Jump and Dressage horses were separated across the PC1 axis (17.18% of the genetic variance). Across the PC2 axis (11.41%), Jump horses were more tightly clustered and Dressage horses were distributed across the axis. The Discovery and Training sets overlapped, with no clustering of one set to the exclusion of the other (Figure S1). Two Jump horses in the Training set clustered more closely with Dressage horses.

3.2. Composite Selection Signals Analysis in Discovery Set

Jump and Dressage were separately analysed as the target cohort in CSS analyses to identify selected ROIs for each discipline. For Jump, 19 clusters of SNPs were identified, 17 of which had ≥ 3 SNPs and were assigned as ROIs. For Dressage, 17 clusters of SNPs were identified, 15 of which had ≥ 3 SNPs and were assigned as ROIs (Table 1).

TABLE 1.

Selection signals for Jump and Dressage.

Chr Cluster SNPs (N) Cluster region (Mb) Max CSS score in cluster Target cohort
17 46 53.89–55.77 2.29 Jump
17 75 21.19–24.18 2.27 Jump
18 37 58.89–60.38 2.21 Jump
28 10 0.79–1.16 2.05 Jump
26 37 30.66–32.17 2.04 Jump
8 59 94.42–97.32 2.03 Jump
21 55 14.18–16.11 1.92 Jump
5 25 37.17–38.28 1.86 Jump
21 46 11.85–13.53 1.83 Jump
4 42 34.13–35.42 1.82 Jump
8 25 60.22–60.99 1.77 Jump
23 16 47.64–48.34 1.73 Jump
1 56 7.7–9.58 1.70 Jump
11 20 61.25–61.65 1.69 Jump
12 7 36.43–36.7 1.65 Jump
2 3 71.81–71.86 1.59 Jump
25 3 19.86–19.93 1.59 Jump
1 1 39.37–39.37 1.56 Jump
12 1 5.88–5.88 1.56 Jump
4 146 14.35–20.45 3.87 Dressage
7 69 56.99–59.81 2.51 Dressage
1 75 19.91–22.73 2.43 Dressage
22 46 16.02–18.45 2.27 Dressage
27 26 39.27–40.23 2.19 Dressage
7 37 2.37–3.75 2.18 Dressage
14 25 36.64–37.56 2.08 Dressage
7 38 52.85–54.52 2.01 Dressage
9 10 46.22–46.72 1.87 Dressage
22 12 33.57–34.02 1.85 Dressage
18 16 21.01–21.59 1.85 Dressage
18 24 69.39–70.17 1.81 Dressage
18 12 27.24–28.02 1.78 Dressage
6 21 8.16–9.12 1.77 Dressage
1 6 13.55–13.62 1.75 Dressage
7 1 62.83–62.83 1.74 Dressage
19 1 60.27–60.27 1.74 Dressage
Open in a new tab

Note: Regions of interest (ROI, Cluster Region) were identified as clusters of ≥ 3 SNPs (Cluster SNPs) among the top 1% SNPs in the composite selections signals (CSS) test.

3.3. Identification and Validation of Lead SNPs

Tests of allelic association were performed for 1127 SNPs within the ROIs to identify the set of lead SNPs. Thirty SNPs were significantly associated with discipline (p < 0.05) (Table S2) in the Discovery set. The 15 SNPs associated with Dressage were different to the 15 SNPs associated with Jump. Accounting for multiple tests, seven SNPs (four Dressage, three Jump) were highly significantly (p < 4.44 × 10−5) associated with discipline. Of the 30 SNPs, 19 were significantly (p < 0.05) associated with discipline in the Training set (Table 2). Following correction for multiple testing, seven SNPs (five Dressage, two Jump) were highly significantly (p < 0.002) associated with discipline. Candidate genes were identified as the closest gene to the lead SNP in each ROI that had a biological function relevant to exercise phenotypes (Table 2). For jumping, five SNPs were intronic variants and eight were < 160 kb from the candidate gene (Table S3). For dressage, three SNPs were intronic variants and eight were < 100 kb from the candidate gene (Table S4).

TABLE 2.

Allelic association test results for the Training set.

CHR SNP A1 Freq. jump Freq. dressage p OR Candidate gene Target discipline
7 rs68747553 T 0.2241 0.6538 5.40E−06 6.5380 GRM5** Dressage
9 rs68871145 G 0.2931 0.6731 6.74E−05 4.9650 VPS13B* Dressage
4 rs69615889 C 0.4310 0.0962 8.24E−05 0.1404 VWC2** Dressage
22 rs1146019687 A 0.2759 0.5962 0.00070 3.8750 PROKR2** Dressage
6 rs68630210 A 0.5000 0.1923 0.00070 0.2381 CYP27A1* Dressage
17 rs69129473 G 0.5385 0.2414 0.00076 0.2727 LPAR6* Jump
17 rs69138655 T 0.6538 0.3621 0.00137 0.3005 SLITRK1** Jump
22 rs69271163 T 0.2241 0.4808 0.00225 3.2050 SRSF6 Dressage
18 rs69133508 A 0.5517 0.2885 0.00471 0.3294 TMEFF2 Dressage
1 rs68587568 G 0.2069 0.0385 0.00533 0.1533 ADRA2A* Dressage
14 rs68951535 A 0.2586 0.5000 0.00814 2.8670 LRRTM2 Dressage
1 rs68480736 C 0.1538 0.0172 0.00895 0.0965 CHST15 Jump
28 rs69415168 A 0.4423 0.2241 0.00906 0.3643 TPH2 Jump
11 rs68923972 T 0.3462 0.5690 0.01491 2.4930 MAP2K3 Jump
21 rs1149211893 G 0.3654 0.1724 0.01931 0.3618 DEPDC1B Jump
18 rs69187435 G 0.1897 0.3846 0.02183 2.6700 ZEB2* Dressage
8 rs68675347 T 0.1923 0.3793 0.02326 2.5670 TPGS2 Jump
7 rs68661332 G 0.5000 0.3077 0.03109 0.4444 MYDGF* Dressage
26 rs396735036 T 0.5769 0.3966 0.04055 0.4819 KCNE1 Jump
Open in a new tab

Note: Candidate genes closest to the lead SNP were identified. *< 0.0001 in Discovery set, **< 0.00001 in Discovery set. p‐value cut‐off following multiple correction p < 0.002, Nominal p‐value cut off p < 0.05.

Abbreviations: A1, minor allele; OR, allelic odds ratio.

The ROIs identified in this study were compared to previously reported QTLs in other warmblood populations. Seven of the 19 Jump ROIs overlapped with previously reported genomic regions, but no overlaps were identified for the Dressage ROIs (Table 3).

TABLE 3.

Overlapping ROIs (this study) with runs of homozygosity (ROH) and copy number variants (CNV) identified in other European warmblood breeds (Nolte et al. 2019; Sole et al. 2019).

Chr ROI (Mb) Overlapping ROH Overlapping CNV Candidate gene in ROI Discipline target
17 21.19–24.18 HOLST HAN TRAKH LPAR6 Jump
18 58.89–60.38 HOLST HAN OLD ITGA4 Jump
21 14.18–16.11 HOLST PLK2 Jump
4 34.13–35.42 TRAKH BEL CDK14 Jump
5 37.17–38.28 TRAKH NES Jump
11 61.25–61.65 BEL MAP2K3 Jump
21 11.85–13.53 BEL DEPDC1B Jump
Open in a new tab

Abbreviations: BEL, Belgian warmblood; HAN, Hanoverian; HOLST, Holsteiner; OLD, Oldenburger; TRAKH, Trakhener.

3.4. PPS Development in Training Set

PPS models were developed for the Training set using five different SNP sets based on different p‐value cut‐offs and weighted using the odds ratio (Discovery set). For all models, there was a significant difference in PPS between disciplines, and the AUC = 0.89 for all models. The model including SNPs p < 0.05 was identified as optimal (p = 2.35 × 10−8, AUC = 0.89) as it had the lowest p‐value (Table S5) and was used for further analyses.

Examination of the PPS distribution among Jump and Dressage horses (Training set) demonstrated clear discrimination of discipline (Table 4, Figure 1). The median PPS for Jump was 6.529 and 6.255, and for Dressage was 1.545 and 2.315 in the Discovery and Training sets, respectively. Using the median (4.704) PPS to predict Dressage (< median) or Jump (> median) had an accuracy of 80% (95% CI 67.03%–89.57%) (Tables S6 and S7).

TABLE 4.

Summary of median PPS in each cohort.

Research set Discipline Median PPS N
Discovery Jump 6.529 21
Discovery Dressage 1.545 17
Training Jump 6.255 28
Training Dressage 2.315 25
Open in a new tab

FIGURE 1.

FIGURE 1

Open in a new tab

PPS distribution among Dressage (light blue) and Jump (dark blue) horses in the Training set.

Jump and Dressage horses across the ranges of PPS categories were as follows: 93% of horses in the D Pro category were Dressage horses, and all horses in the J Pro category were Jump horses; among Dressage horses, 88.5% had a genetic profile indicative for dressage (D and D Pro); among Jump horses, 72% had a genetic profile indicative for jumping (J and J Pro) (Table 5, Figure 2, Figure S2).

TABLE 5.

Overview of distribution of discipline types within each prediction category.

Predicted category % of horses Distribution of discipline type Generalised distribution
D Pro 27.3 93% were D horses 88.5% of D horses
D 29.1 56% were D horses, 44% were J horses
J 27.3 80% were J horses, 20% were D horses 72% of J horses
J Pro 16.4 100% were J horses
Open in a new tab

FIGURE 2.

FIGURE 2

Open in a new tab

Proportion of Jump (J) and Dressage (D) horses within each prediction category (D Pro, D, J and J Pro).

3.5. Validation of the PPS

Among horses with jumping competition records, the median PPS was higher for elite (6.37) than superior (6.02) and inferior (5.46) horses and the PPS was significantly different between elite and inferior (p = 0.043) horses. Jump prediction was highest (85.7%) among elite horses and lowest among inferior horses (69.4%) (Table 6, Figure S3).

TABLE 6.

Median PPS and prediction of Jump (J − J + J Pro) and Dressage (D − D + D Pro) among horses with jumping competition records.

Cohort Description Median PPS J D Total n J D
Elite 5‐star European warmblood 6.37 12 2 14 0.857 0.143
Superior 2 & 3‐star Irish Sport Horse 6.02 12 3 15 0.800 0.200
Inferior < 1 M Irish Sport Horse 5.46 34 15 49 0.694 0.306
Open in a new tab

3.6. PPS Distribution Among Thoroughbred Horses and Event Horses

Among the Thoroughbred horses, the median PPS (4.093) was similar to the median PPS for sport horses (4.704). Using the median PPS (4.704) as previously, 29.1% of Thoroughbreds were predicted to be better suited to jumping and 70.9% better suited to dressage, with 2.1%, 27%, 57.5% and 13.5% having the J Pro, J, D and D Pro profiles respectively (Figure S4). The proportion of J Pro horses (2.1%) was significantly lower (p = 0.0005) than among sport horses (16.4%), but there was no significant difference in the proportion of D Pro horses between Thoroughbreds (13.5%) and sport horses (27.3%). Australia had the highest proportion (2.2%) of Thoroughbreds with the J Pro profile, with the highest proportion (18.6%) with the D Pro profile among European Thoroughbreds.

Among the event horses 57.1% were predicted to be better suited to jumping (J Pro—28.6%, J—28.6%), and 42.8% were predicted to be better suited to dressage (D—28.6%, D Pro—14.3%). The median PPS for event horses (5.635) was higher than the median PPS for Thoroughbreds (4.093) and more event horses (57.1%) were predicted to be better suited to jumping than Thoroughbreds (29.1%). Compared to the jump horses with competition records, the median PPS for event horses was higher than Inferior and lower than Superior horses.

4. Discussion

The genetic improvement of horse breeds requires accurate genetic evaluation of traits of value in order to identify parents with the highest genetic merit for breeding. Traditionally this has been performed using pedigree‐based analysis combined with linear profiling. Although linear profiling provides a system to determine how well a horse conforms to the breeding goal, conformation and movement traits have limited ability to distinguish dressage from jumping horses (Borowska and Lewczuk 2023). As such, more reliable objective predictions are needed. With the recent application of SNP array derived genotypes for parentage verification at registration, most warmblood stud books now systematically calculate genomic estimated breeding values (GEBVs), the aggregate additive genetic value of which half is passed on to offspring (Lande and Thompson 1990; Meuwissen et al. 2001). The GEBV is a linear function of the SNP genotypes of an individual weighted by effect size and assumes contributions to traits from linked loci. A key distinction between GEBVs and polygenic risk models (PRS), referred to here as a polygenic prediction model (PPS), is that a PRS predicts the intrinsic risk (or for PPS, the potential) of an individual for a given trait, whereas GEBVs predict the average value of an individual's parents' genetics that have been inherited (Wray et al. 2019). As a PPS estimates an individual's likelihood of potential for a trait, characterising genetic markers associated with discipline and combining them in a PPS model provides a novel opportunity to identify horses best suited to either jumping or dressage. The PPS could contribute to increasing the accuracy of genetic evaluation for young horses, and as such may be best applied to young animals (foals, yearlings etc.) during early selection processes before they have a recorded phenotype (i.e., via linear profiling, competition etc.). This would enable earlier targeted management and training of animals for the discipline that they are best suited to. Furthermore, specific loci could be used to inform genome‐enabled breeding strategies.

4.1. Development of a PPS for Sporting Discipline

Selection signals approaches are now widely used to identify genes contributing to traits of interest (Schwarzenbacher et al. 2012) as they indicate the presence of an advantageous allele that has increased in frequency due to positive selection (Hoekstra et al. 2001; Grossman et al. 2010). Using this approach, in this study a panel of 30 SNPs with large effect sizes (OR = 2.07–10.36) was catalogued for jumping and dressage. For complex traits such as equine athletic performance, very many genes are likely to be targeted by artificial selection (Han et al. 2022). Such polygenic traits will have contributions from numerous genes influencing the favoured phenotype and these genes may be subject to gene–gene and gene–environment interactions. For complex traits, genetic variants associated with a trait may be combined in prediction models that produce polygenic risk scores (PRS) (Choi et al. 2020). For humans, PRS are used in clinical medicine (Lewis and Vassos 2020), and in the context of athletes, are used to identify individuals that perform more physical activity than others (Kujala et al. 2020), and to indicate power or endurance athletic potential (Jones et al. 2016; Pranckeviciene et al. 2021).

Here, to distinguish between jumping and dressage horses we identified SNPs within or proximal to genes with known biological functions in, and associations with behaviour, brain and nervous system, body composition and growth, bone, cartilage and tendon development, cardiac function and muscle development. The PPS derived from these SNPs had an accuracy of AUC = 0.89 in the Training set, indicating very good to excellent predictive performance of the test (Corbacioglu and Aksel 2023). By assigning the median PPS as the cut‐off value, the accuracy was 80% (95% CI 67.03%–89.57%). The Discovery set comprised only males, with both males and females represented in the Training set. For the Training set, among Jump horses, six horses were incorrectly assigned, five of which were female. Among Dressage horses, five were incorrectly assigned, three of which were female and two male. Therefore, the PPS may be better for prediction among males, particularly for jumping.

Although the prediction accuracy in the Training set was not perfect, this was not unexpected, as the phenotypes were assigned based on pedigree and physical type at a young age (< 2 years old), and not on individual assessment for potential in each discipline (i.e., linear profiling) or competition performance. Therefore, some of the discipline designations to assign the phenotypes may not themselves have been correct. Supporting this, two Jump horses in the Training set clustered more closely with Dressage horses in PC1 of the PCA plot (Figure S1). Notwithstanding this, the accuracy to predict discipline in sport horses using the PPS is considerably higher than the model used to predict power or endurance status among human athletes (66.3%) (Pranckeviciene et al. 2021).

To accurately determine the effectiveness of a polygenic prediction model, out of sample testing must be performed (Choi et al. 2020). Here, among showjumpers variable for performance competition level (Validation set), the PPS was significantly higher among elite than inferior performers (p = 0.043), with the Jump phenotype predicted for 85.7% of elite and 69.4% of inferior horses. Therefore, these results indicate that the test will be valuable for genomic selection to identify horses at an early age that are most suited for jumping. Horses with dressage competition records were not available for the study, and therefore validation of the PPS for dressage competition level will be required. Further validation of the PPS in a larger cohort of warmblood horses divergent for jumping and dressage will be beneficial to provide evidence for the efficacy of testing on a population‐wide scale.

4.2. Overlapping Selection Signals With Other Studies

Strong support for the PPS comes from overlaps with previously reported selection signals in warmbloods providing confidence in the repeatability among sport horses originating from different European warmblood studbooks. Interestingly, there were several common genomic regions that overlapped for jumping (Table 3), but no overlaps were identified for the dressage ROIs. It is unclear why overlaps for the dressage ROIs were not detected; this may be due to limitations due to SNP marker coverage on the lower density array used here than other studies, or may be due to stronger selection for jumping in the stud books previously examined that could obscure dressage signals, or may be due to the different statistical methodology applied. Most of the selection signals reported in the previous studies were identified using a single statistical method. The composite selection signals (CSS) analysis approach applied in this study combines the results of three different statistics into one composite statistic (Randhawa et al. 2014, 2015), providing greater power to detect regions of the genome divergent in their genetic architecture compared to individual constituent tests because each constituent test will be informative on the different characteristics that have shaped the variation at a genomic region (Grossman et al. 2010; Schwarzenbacher et al. 2012).

Among the overlaps for jumping, the ROI that contained LPAR6 overlapped with a selection signal among Swedish warmblood horses identified using the XP‐EHH test (Ablondi et al. 2019), and with a run of homozygosity (ROH) and a selection signal in Holsteiner identified using the iHS test (Nolte et al. 2019). The ROI containing ITGA4 also overlapped with a ROH in Holsteiner that was almost identical in length (Nolte et al. 2019). Furthermore, three ROIs overlapped with copy number variants identified in the Belgian warmblood (Sole et al. 2019), which has consistently ranked among the highest stud books for jumping (Doyle et al. 2022). Although several GWAS studies have identified quantitative trait loci (QTL) for jumping and jumping‐related traits (Schroder et al. 2012; Brard and Ricard 2015; Nazari‐Ghadikolaei et al. 2025), there were no overlapping QTLs among those studies with ROIs in this study.

4.3. Candidate Genes for Jumping

Further support for the PPS arises from the presence of genes in close proximity to the associated SNPs with known biological functions (in other species) highly relevant to the sporting disciplines. For jumping, the associated genes have functions in articular cartilage and tendon development and repair (CHST15, ITGA4, NES) (Luo et al. 2014; Jakobsen et al. 2021; Momen et al. 2023), behaviour and modulation of neuropathic pain (TPH2, SLITRK1, LPAR6) (Waider et al. 2011; Strekalova et al. 2021; Li et al. 2022; Chu et al. 2024; Fan et al. 2024), bone development (CDK14) (Bao et al. 2017; Zhou et al. 2024), cardiac function (KCNE1, PLK2) (Finley et al. 2002; Pedersen et al. 2017), fat deposition (HRAS) (Sevillano et al. 2018; Gozalo‐Marcilla et al. 2021; Duo et al. 2023), muscle (DEPDC1B, MAP2K3) (Figeac et al. 2020; Maasar et al. 2021), and learning and memory (GALR1, MAP2K3) (Miller 1998; Ogren et al. 1998; Rustay et al. 2005; Huentelman et al. 2018).

Examination of genotypes for the two highly significant SNPs for jumping (associated with SLITRK1 and LPAR6) showed considerable variation in frequencies between Dressage and Jump horses. For SLITRK1, 41.4% of Jump horses were homozygous for the favourable allele, compared to 7.7% of Dressage horses. For LPAR6, 55.2% of Jump horses were homozygous for the favourable allele, compared to 20.8% of Dressage horses. SLITRK1 and LPAR6 both function in the modulation of neuropathic pain (Katayama et al. 2010; Chu et al. 2024; Fan et al. 2024), defined as a pathological pain caused by neurological disorders or an injury lesion (Bouhassira and Attal 2011). Neuropathic pain attributed to sport‐induced injuries is common (AlMakadma et al. 2022) and is a possible source of pain in tendinopathies (Webborn 2008). There is a strong link between pain and emotional abnormalities such as anxiety and depression (Michaelides and Zis 2019; Kremer et al. 2021; Damci et al. 2022), and in horses neuropathic pain has been established as a cause of dangerous behaviour (Story et al. 2021). As well as being an animal welfare concern, chronic pain can lead to adverse behaviours that can be dangerous to handlers and riders (de Grauw and van Loon 2016; Torcivia and McDonnell 2021).

In Slitrk1 knockouts, mice display elevated anxiety (Katayama et al. 2010) and in the amygdala, a region of the brain that plays a central role in pain, fear, anxiety and depression, overexpression of the SLITRK1 protein eases neuropathic pain (Chu et al. 2024). In a mouse model simulating nerve‐injury induced neuropathic pain, LPAR6 was found to modulate mechanical and thermal pain thresholds (Fan et al. 2024). Both SLITRK1 and LPAR6 have been suggested as therapeutic targets for alleviating neuropathic pain (Chu et al. 2024; Fan et al. 2024). In several studies investigating selection signals in Thoroughbred horses (Fawcett et al. 2019; McGivney et al. 2020; Han et al. 2022), the most strongly selected region contained ZWINT that also mediates negative behaviour induced by neuropathic pain (Han et al. 2014). Showjumping horses are at increased risk of foot, tarsal and back pain (Murray 2014), and tendon injuries are very common (Smith et al. 2014; Smith and McIlwraith 2021). Therefore, we speculate that genetic variation in SLITRK1 and LPAR6 contributes to the modulation of neuropathic pain in horses that may have a requirement for higher pain thresholds. Studies have shown that adverse behaviours arising from mental or physical discomfort are common in elite showjumpers and dressage horses (Górecka‐Bruzda et al. 2015). Therefore, application of these genetic markers for genetic improvement programmes may also contribute to sport horse welfare.

Nutritional supplements are extremely popular for the treatment of common ailments in sport horses (Agar et al. 2016). For jumping, we identified two genes that encode proteins that are involved in the production of commonly used compounds in nutritional supplements for competition horses. Chondroitin sulphate is used for the treatment of osteoarthritis in humans (Brito et al. 2023), and although the efficacy of some commercial equine supplement products has been questioned (Pearson and Lindinger 2009), it is widely used to support joint health and the maintenance and support of cartilage (Dechant et al. 2005; Agar et al. 2016). CHST15 encodes carbohydrate sulfotransferase 15 that is involved in the early stages of chondroitin sulphate glycosaminoglycan biosynthesis in cartilage (Venkatachalam 2003; Wang et al. 2025). In humans, patients with Kashin‐Beck disease, which causes chronic joint pain and stiffness, CHST15 expression was significantly lower in articular cartilage than in unaffected controls, leading to a reduced ability of the tissue to resist mechanical loads during joint articulation (Luo et al. 2014). The CHST15 SNP had the largest effect size (OR = 10.36) among the Jump SNPs with the favourable allele at a frequency > 98% in Jump compared to 85% in Dressage horses, and 96.6% and 69.2% of Jump and Dressage horses, respectively, were homozygous.

An intronic variant in TPH2, encoding tryptophan hydroxylase 2, which catalyzes the rate‐limiting step in the biosynthesis of serotonin, and modulates anxiety‐related behaviours and emotional regulation (Waider et al. 2011; Strekalova et al. 2021; Li et al. 2022) was also identified for jumping. ‘Calmer’ supplements for horses commonly contain tryptophan to treat excitable horses, but the efficacy can vary (Grimmett and Sillence 2005), suggesting there may be a genetic basis to response to treatment. Based on evidence from rodent studies (Strekalova et al. 2021), it is likely that TPH2 influences resilience to stress and responsiveness to tryptophan supplementation in sport horses. The frequency of the advantageous allele for jumping was 77.6% in Jump horses and 55.8% in Dressage horses, with 3.4% of Jump horses homozygous for the unfavourable genotype compared to 23.1% of Dressage horses. This SNP may be useful to indicate horses that could benefit from tryptophan supplementation for behavioural regulation.

4.4. Candidate Genes for Dressage

Among the 30 SNPs, the most strongly associated SNP, located < 50 kb from GRM5, was identified for dressage. 84.6% of Dressage horses and 41.4% of Jump horses were either homozygous for the favourable allele or heterozygous, representing a genotypic odds ratio of 7.8. GRM5 encodes an excitatory G‐protein coupled receptor that regulates brain function and promotes spatial learning and memory (Balschun et al. 2006). In animals, GRM5 is associated with novelty‐seeking and exploratory behaviours (Parkitna et al. 2013), and modulates locomotor reactivity in a novel environment (Jew et al. 2013). In cattle, GRM5 is the principal gene governing grazing behaviours (Moreno Garcia et al. 2024), and among free ranging cattle is associated with tortuosity of movement (Moreno Garcia et al. 2022), the departure from the path of straightness.

Dressage requires riders and horses to perform in harmony a series of pre‐determined movements within a defined space. Movements include riding in straight lines, lateral movements, geometric figures, and changes of direction performed in different gaits. In some movements (lateral movements), the horse moves in an orientation such that the direction of movement is not aligned with the position of the long axis of the horse's body (de Cocq et al. 2010). Regardless of the movement, according to the FEI Dressage Judging Manual, the horse must remain ‘absolutely straight in any movement on a straight line and bending accordingly when moving on curved lines’ (FEI 2024). Tortuosity of movement relates to an animal's behaviour moving in a path, which has been most widely studied in free ranging foraging animals with freedom of movement, and in enclosed spaces is impacted by the size and the shape of the bounded space (Miller et al. 2011). In the context of the known effect on locomotory behaviour and movement tortuosity in cattle in particular, the identification of GRM5 here indicates it as a major gene influencing the ability of horses to fulfil the requirement for straightness in dressage.

The SNP ranked 2nd in the Discovery set and 4th in the Training set, was located < 50 kb from PROKR2. Homozygotes for the favourable allele and heterozygotes accounted for 88.5% of Dressage and 37.9% of Jump horses, representing a genotypic odds ratio of 12.5. PROKR2 is uniquely expressed in laminae II of the spinal cord, a layer of cells that modulates sensory inputs relayed to the brain from peripheral receptors, and spinal excitatory neurons expressing PROKR2 have been demonstrated to be central to the reward‐based response to pleasant touch (Liu et al. 2022). During a dressage test, the rider must communicate the direction and speed of travel for each movement to the horse through a set of ‘aids’, which include contact with the mouth through the reins, shifting of weight distribution in the saddle, and pressure from the legs on the sides of the horse (de Cocq et al. 2010; Hobbs et al. 2023). The harmony between rider and horse has been described as the visible‐invisible, as each aid given by the rider is almost imperceptible but is visible in the correct execution of the movement by the horse (Clarke 2020). Tactile communication among horses is necessary for maternal and social bonding, reflected in the human–horse relationship by communicating with horses through non‐verbal cues including movement, touch, and space (Argent 2012). Here, we propose that variation in PROKR2, encoding the prokineticin 2 (PROK2) receptor, impacts on the responsiveness of a horse to touch, including subtle pressure cues, especially from the rider's legs. The effect of PROKR2 acts through mechanoreceptors on hairy skin that signal to PROK2, a major neuropeptide in the spinal circuit, directly conveying pleasant touch sensation within the spinal circuit, influencing mice to respond in a reward‐dependent manner to light stroking and punctate stimulation (Liu et al. 2022). The training of horses for dressage requires animals to positively respond to light touch (from the rider's legs) which is rewarded with a withdrawal of the aid once the movement is executed as required. A feature of PROKR2 neurons is that they fatigue with repetitive stimulation (Liu et al. 2022), an uncannily similar description to the situation whereby horses become unresponsive (‘dead to the leg’) if the pressure is not intermittently removed. While it is not possible to directly dissect the exact mechanism by which PROKR2 genetic variation impacts on the dressage horse, we hypothesise that because the neural circuit mechanisms for detecting pleasant touch are highly conserved among species, dressage horses have PROK2‐PROKR2 neurons that are more sensitive to subtle pressure cues from which they are rewarded.

Three other SNPs were highly significantly associated with dressage and were located within and proximal to CYP27A1, VPS13B and VWC2. CYP27A1 and VWC2 are involved in bone formation and development (Almehmadi et al. 2018; Fang et al. 2023), and VPS13B is associated with obesity in humans and body traits in cattle (Farooqi 2005; Jiang et al. 2019). Almost half of the genes for dressage had roles in bone development (VWC2, THSD7B, CYP27A1) and growth traits (VPS13B, TMEFF2, NANOS1, ZEB2) that may contribute to the body size variation observed between dressage and jumping horses. Although conformation and movement traits have limited ability to distinguish dressage from jumping horses (Borowska and Lewczuk 2023), a larger body mass, especially in relation to chest circumference, has been identified in dressage horses compared to jumping horses (Pagan et al. 2009). PRKAG3 is the closest gene to CYP27A1. In the horse, a highly conserved variant in PRKAG3 occurs only in heavy and moderately heavy breeds, but not in light horse breeds or ponies (Park et al. 2003). Knockout experiments in mice have demonstrated that AMPK γ3 (encoded by PRKAG3) influences enhanced work performance and reduces fatigue in skeletal muscle (Nilsson et al. 2006). Given that fatigue in dressage horses can influence stride characteristics and fluidity of movement (Pasquiet et al. 2022), the SNP identified here may be linked to genetic variation in PRKAG3 which could influence tolerance to muscle fatigue, thus contributing to the maintenance of carriage of the shape and correctness of movement even in fatigued horses.

4.5. Usefulness of the PPS for Identifying Second Careers for Thoroughbred Racehorses

To examine the potential for across breed application of the genetic markers identified here, we examined the distribution of the PPS in a sample of Thoroughbred horses. The rehoming and retraining of Thoroughbred horses following retirement from racing is a key goal to improve welfare efforts in the Thoroughbred industry (Stowe and Kibler 2016; Crawford et al. 2021). Both behavioural and genetic information have been proposed to assist with identifying suitable sporting disciplines for retired Thoroughbreds (Ryu et al. 2024), with a bioinformatics‐based approach identifying several behaviour‐related genes with polymorphisms that could be useful to select second careers (Yokomori et al. 2023). As well as temperament and behavioural considerations, understanding the genetic potential for a horse to be better suited to a second career in jumping or dressage could be beneficial towards this goal. The distribution of the PPS among Thoroughbreds indicates that second careers in dressage, eventing or leisure riding are more suitable for Thoroughbreds than jumping; however, 29.1% of Thoroughbreds demonstrated a PPS suitable for jumping. Jumping is a preferred characteristic for rehoming Thoroughbreds, with horses deemed to be suitable for jumping being rehomed 50% faster than others (Stowe and Kibler 2016). Despite this, horses considered sound for jumping were more likely to be returned post‐adoption, which suggested that owners had expectations of the horse for jumping that were not met (Stowe and Kibler 2016). Using the PPS could identify up to 2.1% of Thoroughbreds with the best aptitude for jumping which could target these horses for more competitive riders.

Many Thoroughbreds have second careers in eventing. Among the event horses, more were suited to jumping than Thoroughbreds, and the event horses had a median PPS between the Inferior and Superior jump horses. Although jumping is a critical trait for event horses, they are not required to jump at the same level as elite showjumpers, and depending on the competition level and type, jumping and dressage have different contributions. Nonetheless, these results suggest that up to one‐third of Thoroughbreds are genetically suited to jumping and those horses may also be suitable for eventing. Screening horses for suitability for new owner preferences has been suggested to improve the success of the rehoming process (Stowe and Kibler 2016), towards which the PPS could be applied. This work provides a proof of concept that would need to be validated in a sample of Thoroughbreds that have been rehomed and used for different disciplines.

5. Conclusion

We have developed a highly accurate (AUC = 0.89) polygenic prediction model that clearly distinguishes between horses suited to jumping and dressage. The PPS contains SNPs within or proximal to genes with functions relevant to equestrian sports. For jumping, the most strongly associated genes function in the modulation of neuropathic pain (LPAR6 and SLITRK1), and for dressage, the most strongly associated genes function in tortuosity of movement (GRM5) and responsiveness to touch (PROKR2). The validation of the PPS in an independent set of proven jumping horses demonstrates the direct applicability of the model for identifying horses suited to jumping. Based on the strength of the SNP associations for dressage, and the candidate gene functions, it is likely that the prediction of dressage may be stronger than for jumping. However, further work will be required to validate the SNPs for dressage among horses divergent for competition level. The PPS will likely be best applied to young horses during early selection processes before the phenotype is evident via linear profiling or competition results.

Author Contributions

Emmeline W. Hill: conceptualization, investigation, funding acquisition, writing – original draft, methodology, validation, formal analysis, project administration, supervision, resources. Heinrich Anhold: investigation, resources, validation. Haige Han: methodology, formal analysis, writing – review and editing, data curation. Amy R. Holtby: investigation, writing – review and editing, data curation. Beatrice A. McGivney: investigation, writing – review and editing, methodology, formal analysis, supervision.

Conflicts of Interest

E.W.H. is a director, and H.H., A.R.H. and B.A.M. are employees of Zinto Labs that funded the research. H.A. is a director of Epona Biotech that contributed to the research. Other than the researchers, there was no other input from the companies.

Supporting information

Table S1: Details for the discovery, training and validation sets.

Table S2: Allelic association test results for the Discovery set. The top SNP (lowest p‐value) in each ROI is shown. SNPs are shown ranked by p‐value for association with discipline.

Table S3: Candidate genes for jumping. Candidate genes identified based on proximity to the lead SNP, and biological function relevant to jumping. Highly significant SNPs are in bold.

Table S4: Candidate genes for dressage. Candidate genes identified based on proximity to the lead SNP, and biological function relevant to dressage. Highly significant SNPs are in bold.

Table S5: Summary statistics for PPS models using different SNP sets for a range of p‐value cut‐offs. The p‐value shown is the result of a t‐test comparing the PPS between Jump and Dressage horses in the Training set.

Table S6: Confusion matrix showing how many horses were correctly and incorrectly assigned to each discipline using the median PPS as the cut‐off value.

Table S7: Accuracy to predict Jump and Dressage using the median PPS as the cut‐off value.

Figure S1: Principal component analysis (PCA) plot illustrating the distribution of the genetic variation among Jump and Dressage horses for the first two PCs that accounted for 28.6% of the genetic variance.

Figure S2: Proportion of Jump and Dressage horses within each prediction category.

Figure S3: Proportion of horses with jumping competition records categorised as Jump from the PPS.

Figure S4: Genetic predictions of second careers for Thoroughbreds in Australia, Europe and USA, using the PPS developed for sport horses (D Pro—dark blue, D—orange, J—green, J Pro—light blue).

AGE-57-0-s001.docx (155KB, docx)

Acknowledgements

The authors thank Dr. Jörg Aurich and Dr. Christine Aurich for access to the genotype data and phenotype information for the main study horses, and the owners of the horses used for validation.

Data Availability Statement

The data that support the findings of this study are not publicly available due to privacy restrictions. The data may be made available for academic purposes after signing a Material Transfer Agreement, which will be subject to confidentiality and will prevent third parties from commercial exploitation.

References

  1. Ablondi, M. , Eriksson S., Tetu S., Sabbioni A., Viklund A., and Mikko S.. 2019. “Genomic Divergence in Swedish Warmblood Horses Selected for Equestrian Disciplines.” Genes (Basel) 10: 976. [DOI] [PMC free article] [PubMed] [Google Scholar]
  2. Agar, C. , Gemmill R., Hollands T., and Freeman S. L.. 2016. “The Use of Nutritional Supplements in Dressage and Eventing Horses.” Veterinary Record Open 3: e000154. [DOI] [PMC free article] [PubMed] [Google Scholar]
  3. AlMakadma, Y. , Eirale C., and Chamari K.. 2022. “Neuropathic Pain in Athletes: Basics of Diagnosis and Monitoring of a Hidden Threat.” Biology of Sport 39: 943–949. [DOI] [PMC free article] [PubMed] [Google Scholar]
  4. Almehmadi, A. , Ohyama Y., Kaku M., et al. 2018. “VWC2 Increases Bone Formation Through Inhibiting Activin Signaling.” Calcified Tissue International 103: 663–674. [DOI] [PMC free article] [PubMed] [Google Scholar]
  5. Argent, G. 2012. “Toward a Privileging of the Nonverbal: Communication, Corporeal Synchrony, and Transcendence in Humans and Horses.” In Experiencing Animal Minds: An Anthology of Animal‐Human Encounters, edited by Smith J. A. and Mitchell R. W., 111–128. Columbia University Press. [Google Scholar]
  6. Balschun, D. , Zuschratter W., and Wetzel W.. 2006. “Allosteric Enhancement of Metabotropic Glutamate Receptor 5 Function Promotes Spatial Memory.” Neuroscience 142: 691–702. [DOI] [PubMed] [Google Scholar]
  7. Bao, Q. , Chen S., Qin H., et al. 2017. “Constitutive Beta‐Catenin Activation in Osteoblasts Impairs Terminal Osteoblast Differentiation and Bone Quality.” Experimental Cell Research 350: 123–131. [DOI] [PubMed] [Google Scholar]
  8. Bonow, S. , Eriksson S., Thoren Hellsten E., and Gelinder Viklund A.. 2023. “Consequences of Specialized Breeding in the Swedish Warmblood Horse Population.” Journal of Animal Breeding and Genetics 140: 79–91. [DOI] [PMC free article] [PubMed] [Google Scholar]
  9. Borowska, A. , and Lewczuk D.. 2023. “Comparison of Conformation and Movement Characteristics in Dressage and Jumping Sport Warmblood Mares Based on Point Evaluation and Linear Scoring System.” Animals (Basel) 13: 3101. [DOI] [PMC free article] [PubMed] [Google Scholar]
  10. Bouhassira, D. , and Attal N.. 2011. “Diagnosis and Assessment of Neuropathic Pain: The Saga of Clinical Tools.” Pain 152: S74–S83. [DOI] [PubMed] [Google Scholar]
  11. Brard, S. , and Ricard A.. 2015. “Genome‐Wide Association Study for Jumping Performances in French Sport Horses.” Animal Genetics 46: 78–81. [DOI] [PubMed] [Google Scholar]
  12. Brito, R. , Costa D., Dias C., Cruz P., and Barros P.. 2023. “Chondroitin Sulfate Supplements for Osteoarthritis: A Critical Review.” Cureus 15: e40192. [DOI] [PMC free article] [PubMed] [Google Scholar]
  13. Chang, C. C. , Chow C. C., Tellier L. C., Vattikuti S., Purcell S. M., and Lee J. J.. 2015. “Second‐Generation PLINK: Rising to the Challenge of Larger and Richer Datasets.” GigaScience 4: 7. [DOI] [PMC free article] [PubMed] [Google Scholar]
  14. Choi, S. W. , Mak T. S., and O'Reilly P. F.. 2020. “Tutorial: A Guide to Performing Polygenic Risk Score Analyses.” Nature Protocols 15: 2759–2772. [DOI] [PMC free article] [PubMed] [Google Scholar]
  15. Chu, R. , Lu Y., Fan X., et al. 2024. “Changes in SLITRK1 Level in the Amygdala Mediate Chronic Neuropathic Pain‐Induced Anxio‐Depressive Behaviors in Mice.” Journal of Integrative Neuroscience 23: 82. [DOI] [PubMed] [Google Scholar]
  16. Clarke, P. N. 2020. “Nursing Theory Informs Dressage.” Nursing Science Quarterly 33: 35–37. [DOI] [PubMed] [Google Scholar]
  17. Corbacioglu, S. K. , and Aksel G.. 2023. “Receiver Operating Characteristic Curve Analysis in Diagnostic Accuracy Studies: A Guide to Interpreting the Area Under the Curve Value.” Turkish Journal of Emergency Medicine 23: 195–198. [DOI] [PMC free article] [PubMed] [Google Scholar]
  18. Crawford, K. L. , Finnane A., Greer R. M., et al. 2021. “Appraising the Welfare of Thoroughbred Racehorses in Training in Queensland, Australia: The Incidence, Risk Factors and Outcomes for Horses After Retirement From Racing.” Animals (Basel) 11: 142. [DOI] [PMC free article] [PubMed] [Google Scholar]
  19. Damci, A. , Schruers K. R. J., Leue C., Faber C. G., and Hoeijmakers J. G. J.. 2022. “Anxiety and Depression in Small Fiber Neuropathy.” Journal of the Peripheral Nervous System 27: 291–301. [DOI] [PMC free article] [PubMed] [Google Scholar]
  20. de Cocq, P. , Mooren M., Dortmans A., et al. 2010. “Saddle and Leg Forces During Lateral Movements in Dressage.” Equine Veterinary Journal. Supplement 42: 644–649. [DOI] [PubMed] [Google Scholar]
  21. de Grauw, J. C. , and van Loon J. P.. 2016. “Systematic Pain Assessment in Horses.” Veterinary Journal 209: 14–22. [DOI] [PubMed] [Google Scholar]
  22. Dechant, J. E. , Baxter G. M., Frisbie D. D., Trotter G. W., and McIlwraith C. W.. 2005. “Effects of Glucosamine Hydrochloride and Chondroitin Sulphate, Alone and in Combination, on Normal and Interleukin‐1 Conditioned Equine Articular Cartilage Explant Metabolism.” Equine Veterinary Journal 37: 227–231. [DOI] [PubMed] [Google Scholar]
  23. Doyle, J. L. , Carroll C. J., Corbally A. F., and Fahey A. G.. 2022. “An Overview of International Genetic Evaluations of Show Jumping in Sport Horses.” Translational Animal Science 6: txac038. [DOI] [PMC free article] [PubMed] [Google Scholar]
  24. Duo, T. , Liu X., Mo D., et al. 2023. “Single‐Base Editing in IGF2 Improves Meat Production and Intramuscular Fat Deposition in Liang Guang Small Spotted Pigs.” Journal of Animal Science and Biotechnology 14: 141. [DOI] [PMC free article] [PubMed] [Google Scholar]
  25. Fan, X. , Chu R., Jiang X., et al. 2024. “LPAR6 Participates in Neuropathic Pain by Mediating Astrocyte Cells via ROCK2/NF‐KappaB Signal Pathway.” Molecular Neurobiology 61: 8402–8413. [DOI] [PubMed] [Google Scholar]
  26. Fang, Z. , Cheng G., He M., and Lin Y.. 2023. “CYP27A1 Deficiency Promoted Osteoclast Differentiation.” PeerJ 11: e15041. [DOI] [PMC free article] [PubMed] [Google Scholar]
  27. Farooqi, I. S. 2005. “Genetic and Hereditary Aspects of Childhood Obesity.” Best Practice & Research. Clinical Endocrinology & Metabolism 19: 359–374. [DOI] [PubMed] [Google Scholar]
  28. Fawcett, J. A. , Sato F., Sakamoto T., Iwasaki W. M., Tozaki T., and Innan H.. 2019. “Genome‐Wide SNP Analysis of Japanese Thoroughbred Racehorses.” PLoS One 14: e0218407. [DOI] [PMC free article] [PubMed] [Google Scholar]
  29. FEI . 2024. FEI Dressage Judging Manual. F.E. Internationale. www.fei.org. [Google Scholar]
  30. Figeac, N. , Pruller J., Hofer I., et al. 2020. “DEPDC1B Is a Key Regulator of Myoblast Proliferation in Mouse and Man.” Cell Proliferation 53: e12717. [DOI] [PMC free article] [PubMed] [Google Scholar]
  31. Finley, M. R. , Li Y., Hua F., et al. 2002. “Expression and Coassociation of ERG1, KCNQ1, and KCNE1 Potassium Channel Proteins in Horse Heart.” American Journal of Physiology. Heart and Circulatory Physiology 283: H126–H138. [DOI] [PubMed] [Google Scholar]
  32. Górecka‐Bruzda, A. , Kosińska I., Jaworski Z., Tadeusz Jezierski T., and Murphy J.. 2015. “Conflict Behavior in Elite Show Jumping and Dressage Horses.” Journal of Veterinary Behavior 10: 137–146. [Google Scholar]
  33. Gozalo‐Marcilla, M. , Buntjer J., Johnsson M., et al. 2021. “Genetic Architecture and Major Genes for Backfat Thickness in Pig Lines of Diverse Genetic Backgrounds.” Genetics, Selection, Evolution 53: 76. [DOI] [PMC free article] [PubMed] [Google Scholar]
  34. Grimmett, A. , and Sillence M. N.. 2005. “Calmatives for the Excitable Horse: A Review of L‐Tryptophan.” Veterinary Journal 170: 24–32. [DOI] [PubMed] [Google Scholar]
  35. Grossman, S. R. , Shlyakhter I., Karlsson E. K., et al. 2010. “A Composite of Multiple Signals Distinguishes Causal Variants in Regions of Positive Selection.” Science 327: 883–886. [DOI] [PubMed] [Google Scholar]
  36. Han, H. , McGivney B. A., Allen L., et al. 2022. “Common Protein‐Coding Variants Influence the Racing Phenotype in Galloping Racehorse Breeds.” Communications Biology 5: 1320. [DOI] [PMC free article] [PubMed] [Google Scholar]
  37. Han, M. , Xiao X., Yang Y., et al. 2014. “SIP30 Is Required for Neuropathic Pain‐Evoked Aversion in Rats.” Journal of Neuroscience 34: 346–355. [DOI] [PMC free article] [PubMed] [Google Scholar]
  38. Heuer, C. , Scheel C., Tetens J., Kuhn C., and Thaller G.. 2016. “Genomic Prediction of Unordered Categorical Traits: An Application to Subpopulation Assignment in German Warmblood Horses.” Genetics, Selection, Evolution 48: 13. [DOI] [PMC free article] [PubMed] [Google Scholar]
  39. Hobbs, S. J. , Serra Braganca F. M., Rhodin M., Hernlund E., Peterson M., and Clayton H. M.. 2023. “Evaluating Overall Performance in High‐Level Dressage Horse‐Rider Combinations by Comparing Measurements From Inertial Sensors With General Impression Scores Awarded by Judges.” Animals (Basel) 13: 2496. [DOI] [PMC free article] [PubMed] [Google Scholar]
  40. Hoekstra, H. E. , Hoekstra J. M., Berrigan D., et al. 2001. “Strength and Tempo of Directional Selection in the Wild.” Proceedings of the National Academy of Sciences of the United States of America 98: 9157–9160. [DOI] [PMC free article] [PubMed] [Google Scholar]
  41. Huentelman, M. J. , Piras I. S., Siniard A. L., et al. 2018. “Associations of MAP2K3 Gene Variants With Superior Memory in SuperAgers.” Frontiers in Aging Neuroscience 10: 155. [DOI] [PMC free article] [PubMed] [Google Scholar]
  42. Jakobsen, J. R. , Schjerling P., Svensson R. B., et al. 2021. “RNA Sequencing and Immunofluorescence of the Myotendinous Junction of Mature Horses and Humans.” American Journal of Physiology. Cell Physiology 321: C453–C470. [DOI] [PubMed] [Google Scholar]
  43. Jew, C. P. , Wu C. S., Sun H., et al. 2013. “mGluR5 Ablation in Cortical Glutamatergic Neurons Increases Novelty‐Induced Locomotion.” PLoS One 8: e70415. [DOI] [PMC free article] [PubMed] [Google Scholar]
  44. Jiang, J. , Cole J. B., Freebern E., Da Y., VanRaden P. M., and Ma L.. 2019. “Functional Annotation and Bayesian Fine‐Mapping Reveals Candidate Genes for Important Agronomic Traits in Holstein Bulls.” Communications Biology 2: 212. [DOI] [PMC free article] [PubMed] [Google Scholar]
  45. Jones, N. , Kiely J., Suraci B., et al. 2016. “A Genetic‐Based Algorithm for Personalized Resistance Training.” Biology of Sport 33: 117–126. [DOI] [PMC free article] [PubMed] [Google Scholar]
  46. Katayama, K. , Yamada K., Ornthanalai V. G., et al. 2010. “Slitrk1‐Deficient Mice Display Elevated Anxiety‐Like Behavior and Noradrenergic Abnormalities.” Molecular Psychiatry 15: 177–184. [DOI] [PubMed] [Google Scholar]
  47. Kremer, M. , Becker L. J., Barrot M., and Yalcin I.. 2021. “How to Study Anxiety and Depression in Rodent Models of Chronic Pain?” European Journal of Neuroscience 53: 236–270. [DOI] [PubMed] [Google Scholar]
  48. Kujala, U. M. , Palviainen T., Pesonen P., et al. 2020. “Polygenic Risk Scores and Physical Activity.” Medicine and Science in Sports and Exercise 52: 1518–1524. [DOI] [PMC free article] [PubMed] [Google Scholar]
  49. Lande, R. , and Thompson R.. 1990. “Efficiency of Marker‐Assisted Selection in the Improvement of Quantitative Traits.” Genetics 124: 743–756. [DOI] [PMC free article] [PubMed] [Google Scholar]
  50. Lewis, C. M. , and Vassos E.. 2020. “Polygenic Risk Scores: From Research Tools to Clinical Instruments.” Genome Medicine 12: 44. [DOI] [PMC free article] [PubMed] [Google Scholar]
  51. Li, Y. , Zhu S., Xie K., et al. 2022. “TLR4 in Tph2 Neurons Modulates Anxiety‐Related Behaviors in a Sex‐Dependent Manner.” Neuropharmacology 216: 109175. [DOI] [PubMed] [Google Scholar]
  52. Lindsay‐McGee, V. , Sanchez‐Molano E., Banos G., Clark E. L., Piercy R. J., and Psifidi A.. 2023. “Genetic Characterisation of the Connemara Pony and the Warmblood Horse Using a Within‐Breed Clustering Approach.” Genetics, Selection, Evolution 55: 60. [DOI] [PMC free article] [PubMed] [Google Scholar]
  53. Liu, B. , Qiao L., Liu K., Liu J., Piccinni‐Ash T. J., and Chen Z. F.. 2022. “Molecular and Neural Basis of Pleasant Touch Sensation.” Science 376: 483–491. [DOI] [PMC free article] [PubMed] [Google Scholar]
  54. Luo, M. , Chen J., Li S., et al. 2014. “Changes in the Metabolism of Chondroitin Sulfate Glycosaminoglycans in Articular Cartilage From Patients With Kashin‐Beck Disease.” Osteoarthritis and Cartilage 22: 986–995. [DOI] [PubMed] [Google Scholar]
  55. Maasar, M. F. , Turner D. C., Gorski P. P., et al. 2021. “The Comparative Methylome and Transcriptome After Change of Direction Compared to Straight Line Running Exercise in Human Skeletal Muscle.” Frontiers in Physiology 12: 619447. [DOI] [PMC free article] [PubMed] [Google Scholar]
  56. McGivney, B. A. , Han H., Corduff L. R., et al. 2020. “Genomic Inbreeding Trends, Influential Sire Lines and Selection in the Global Thoroughbred Horse Population.” Scientific Reports 10: 466. [DOI] [PMC free article] [PubMed] [Google Scholar]
  57. McGivney, B. A. , Harty D., Corbally A., and Hill E. W.. 2023. “The Genetic Composition of the Traditional Irish Horse—Towards the Development of a DNA‐Ancestry Test for the Preservation of Traditionally Bred Irish Sport Horses.” Genetic Resources 4: 29–36. [Google Scholar]
  58. Meuwissen, T. H. , Hayes B. J., and Goddard M. E.. 2001. “Prediction of Total Genetic Value Using Genome‐Wide Dense Marker Maps.” Genetics 157: 1819–1829. [DOI] [PMC free article] [PubMed] [Google Scholar]
  59. Michaelides, A. , and Zis P.. 2019. “Depression, Anxiety and Acute Pain: Links and Management Challenges.” Postgraduate Medicine 131: 438–444. [DOI] [PubMed] [Google Scholar]
  60. Miller, C. , Christman M. C., and Estevez I.. 2011. “Movement in a Confined Space: Estimating Path Tortuosity.” Applied Animal Behaviour Science 135: 13–23. [Google Scholar]
  61. Miller, M. A. 1998. “Regulation of Galanin in Memory Pathways.” Annals of the New York Academy of Sciences 863: 323–341. [DOI] [PubMed] [Google Scholar]
  62. Momen, M. , Brauer K., Patterson M. M., et al. 2023. “Genetic Architecture and Polygenic Risk Score Prediction of Degenerative Suspensory Ligament Desmitis (DSLD) in the Peruvian Horse.” Frontiers in Genetics 14: 1201628. [DOI] [PMC free article] [PubMed] [Google Scholar]
  63. Moreno Garcia, C. A. , Perelman S. B., Dynes R., Maxwell T. M. R., Zhou H., and Hickford J.. 2024. “Key Grazing Behaviours of Beef Cattle Identify Specific Genotypes of the Glutamate Metabotropic Receptor 5 Gene (GRM5).” Behavior Genetics 54: 212–229. [DOI] [PMC free article] [PubMed] [Google Scholar]
  64. Moreno Garcia, C. A. , Zhou H., Altimira D., et al. 2022. “The Glutamate Metabotropic Receptor 5 (GRM5) Gene Is Associated With Beef Cattle Home Range and Movement Tortuosity.” Journal of Animal Science and Biotechnology 13: 111. [DOI] [PMC free article] [PubMed] [Google Scholar]
  65. Murray, R. C. 2014. “Veterinary Aspects of Training the Show Jumping Horse.” In Equine Sports Medicine and Surgery, edited by Hinchcliff K. W., Kaneps A. J., and Geor R. J., 1127–1135. Saunders Elsevier. [Google Scholar]
  66. Nazari‐Ghadikolaei, A. , Fikse W. F., Viklund A. G., Mikko S., and Eriksson S.. 2025. “Single‐Step Genome‐Wide Association Study of Factors for Evaluated and Linearly Scored Traits in Swedish Warmblood Horses.” Journal of Animal Breeding and Genetics 142: 499–512. [DOI] [PMC free article] [PubMed] [Google Scholar]
  67. Nilsson, E. C. , Long Y. C., Martinsson S., et al. 2006. “Opposite Transcriptional Regulation in Skeletal Muscle of AMP‐Activated Protein Kinase gamma3 R225Q Transgenic Versus Knock‐Out Mice.” Journal of Biological Chemistry 281: 7244–7252. [DOI] [PubMed] [Google Scholar]
  68. Nolte, W. , Thaller G., and Kuehn C.. 2019. “Selection Signatures in Four German Warmblood Horse Breeds: Tracing Breeding History in the Modern Sport Horse.” PLoS One 14: e0215913. [DOI] [PMC free article] [PubMed] [Google Scholar]
  69. Ogren, S. O. , Schott P. A., Kehr J., et al. 1998. “Modulation of Acetylcholine and Serotonin Transmission by Galanin. Relationship to Spatial and Aversive Learning.” Annals of the New York Academy of Sciences 863: 342–363. [DOI] [PubMed] [Google Scholar]
  70. Pagan, J. D. , Martin O. A., and Crowley N. L.. 2009. “Relationship Between Body Condition and Metabolic Parameters in Sport Horses, Pony Hunters and Polo Ponies.” Journal of Equine Veterinary Science 29: 418–420. [Google Scholar]
  71. Park, H. B. , Marklund S., Jeon J. T., et al. 2003. “Molecular Characterization and Mutational Screening of the PRKAG3 Gene in the Horse.” Cytogenetic and Genome Research 102: 211–216. [DOI] [PubMed] [Google Scholar]
  72. Parkitna, J. R. , Sikora M., Golda S., et al. 2013. “Novelty‐Seeking Behaviors and the Escalation of Alcohol Drinking After Abstinence in Mice Are Controlled by Metabotropic Glutamate Receptor 5 on Neurons Expressing Dopamine d1 Receptors.” Biological Psychiatry 73: 263–270. [DOI] [PubMed] [Google Scholar]
  73. Pasquiet, B. , Biau S., Trebot Q., Debril J. F., Durand F., and Fradet L.. 2022. “Detection of Horse Locomotion Modifications due to Training With Inertial Measurement Units: A Proof‐of‐Concept.” Sensors (Basel) 22: 4981. [DOI] [PMC free article] [PubMed] [Google Scholar]
  74. Pearson, W. , and Lindinger M.. 2009. “Low Quality of Evidence for Glucosamine‐Based Nutraceuticals in Equine Joint Disease: Review of In Vivo Studies.” Equine Veterinary Journal 41: 706–712. [DOI] [PubMed] [Google Scholar]
  75. Pedersen, P. J. , Thomsen K. B., Flak J. B., et al. 2017. “Molecular Cloning and Functional Expression of the K(+) Channel K(V)7.1 and the Regulatory Subunit KCNE1 From Equine Myocardium.” Research in Veterinary Science 113: 79–86. [DOI] [PubMed] [Google Scholar]
  76. Pranckeviciene, E. , Gineviciene V., Jakaitiene A., Januska L., and Utkus A.. 2021. “Total Genotype Score Modelling of Polygenic Endurance‐Power Profiles in Lithuanian Elite Athletes.” Genes (Basel) 12: 1067. [DOI] [PMC free article] [PubMed] [Google Scholar]
  77. Randhawa, I. A. , Khatkar M. S., Thomson P. C., and Raadsma H. W.. 2014. “Composite Selection Signals Can Localize the Trait Specific Genomic Regions in Multi‐Breed Populations of Cattle and Sheep.” BMC Genetics 15: 34. [DOI] [PMC free article] [PubMed] [Google Scholar]
  78. Randhawa, I. A. , Khatkar M. S., Thomson P. C., and Raadsma H. W.. 2015. “Composite Selection Signals for Complex Traits Exemplified Through Bovine Stature Using Multibreed Cohorts of European and African Bos taurus .” G3 (Bethesda) 5: 1391–1401. [DOI] [PMC free article] [PubMed] [Google Scholar]
  79. Rovere, G. , Ducro B. J., van Arendonk J. A., Norberg E., and Madsen P.. 2017. “Genetic Correlations Between Dressage, Show Jumping and Studbook‐Entry Inspection Traits in a Process of Specialization in Dutch Warmblood Horses.” Journal of Animal Breeding and Genetics 134: 162–171. [DOI] [PubMed] [Google Scholar]
  80. Rovere, G. , Madsen P., Norberg E., van Arendonk J. A., and Ducro B. J.. 2015. “Effect of Specialization on Genetic Parameters of Studbook‐Entry Inspection in Dutch Warmblood Horses.” Journal of Animal Breeding and Genetics 132: 441–448. [DOI] [PubMed] [Google Scholar]
  81. Rustay, N. R. , Wrenn C. C., Kinney J. W., et al. 2005. “Galanin Impairs Performance on Learning and Memory Tasks: Findings From Galanin Transgenic and GAL‐R1 Knockout Mice.” Neuropeptides 39: 239–243. [DOI] [PubMed] [Google Scholar]
  82. Ryu, S. H. , Lee K. E., Forbes E., et al. 2024. “Behavioral and Cardiac Responses to a Model Startle Test to Assess Retired Thoroughbred Racehorses for Equestrians.” Journal of Veterinary Science 25: e84. [DOI] [PMC free article] [PubMed] [Google Scholar]
  83. Schroder, W. , Klostermann A., Stock K. F., and Distl O.. 2012. “A Genome‐Wide Association Study for Quantitative Trait Loci of Show‐Jumping in Hanoverian Warmblood Horses.” Animal Genetics 43: 392–400. [DOI] [PubMed] [Google Scholar]
  84. Schwarzenbacher, H. , Dolezal M., Flisikowski K., et al. 2012. “Combining Evidence of Selection With Association Analysis Increases Power to Detect Regions Influencing Complex Traits in Dairy Cattle.” BMC Genomics 13: 48. [DOI] [PMC free article] [PubMed] [Google Scholar]
  85. Sevillano, C. A. , Ten Napel J., Guimaraes S. E. F., Silva F. F., and Calus M. P. L.. 2018. “Effects of Alleles in Crossbred Pigs Estimated for Genomic Prediction Depend on Their Breed‐of‐Origin.” BMC Genomics 19: 740. [DOI] [PMC free article] [PubMed] [Google Scholar]
  86. Smith, R. , McIlwraith W., Schweitzer R., et al. 2014. “Advances in the Understanding of Tendinopathies: A Report on the Second Havemeyer Workshop on Equine Tendon Disease.” Equine Veterinary Journal 46: 4–9. [DOI] [PubMed] [Google Scholar]
  87. Smith, R. K. W. , and McIlwraith C. W.. 2021. “‘One Health’ in Tendinopathy Research: Current Concepts.” Journal of Orthopaedic Research 39: 1596–1602. [DOI] [PubMed] [Google Scholar]
  88. Sole, M. , Ablondi M., Binzer‐Panchal A., et al. 2019. “Inter‐ and Intra‐Breed Genome‐Wide Copy Number Diversity in a Large Cohort of European Equine Breeds.” BMC Genomics 20: 759. [DOI] [PMC free article] [PubMed] [Google Scholar]
  89. Story, M. R. , Nout‐Lomas Y. S., Aboellail T. A., et al. 2021. “Dangerous Behavior and Intractable Axial Skeletal Pain in Performance Horses: A Possible Role for Ganglioneuritis (14 Cases; 2014–2019).” Frontiers in Veterinary Science 8: 734218. [DOI] [PMC free article] [PubMed] [Google Scholar]
  90. Stowe, C. J. , and Kibler M. L.. 2016. “Characteristics of Adopted Thoroughbred Racehorses in Second Careers.” Journal of Applied Animal Welfare Science 19: 81–89. [DOI] [PubMed] [Google Scholar]
  91. Strekalova, T. , Svirin E., Waider J., et al. 2021. “Altered Behaviour, Dopamine and Norepinephrine Regulation in Stressed Mice Heterozygous in TPH2 Gene.” Progress in Neuro‐Psychopharmacology & Biological Psychiatry 108: 110155. [DOI] [PubMed] [Google Scholar]
  92. Torcivia, C. , and McDonnell S.. 2021. “Equine Discomfort Ethogram.” Animals (Basel) 11: 580. [DOI] [PMC free article] [PubMed] [Google Scholar]
  93. Venkatachalam, K. V. 2003. “Human 3′‐Phosphoadenosine 5′‐Phosphosulfate (PAPS) Synthase: Biochemistry, Molecular Biology and Genetic Deficiency.” IUBMB Life 55: 1–11. [DOI] [PubMed] [Google Scholar]
  94. Waider, J. , Araragi N., Gutknecht L., and Lesch K. P.. 2011. “Tryptophan Hydroxylase‐2 (TPH2) in Disorders of Cognitive Control and Emotion Regulation: A Perspective.” Psychoneuroendocrinology 36: 393–405. [DOI] [PubMed] [Google Scholar]
  95. Wang, Z. , Song W., Wei W., et al. 2025. “Structural and Mechanism‐Based Engineering of Sulfotransferase CHST15 for the Efficient Synthesis of Chondroitin Sulfate E.” Applied and Environmental Microbiology 91: e0157324. [DOI] [PMC free article] [PubMed] [Google Scholar]
  96. Webborn, A. D. 2008. “Novel Approaches to Tendinopathy.” Disability and Rehabilitation 30: 1572–1577. [DOI] [PubMed] [Google Scholar]
  97. Welker, V. , Stock K., Schöpke K., and Swalve H. H.. 2018. “Genetic Parameters of New Comprehensive Performance Traits for Dressage and Show Jumping Competitions Performance of German Riding Horses.” Livestock Science 212: 93–98. [Google Scholar]
  98. Wray, N. R. , Kemper K. E., Hayes B. J., Goddard M. E., and Visscher P. M.. 2019. “Complex Trait Prediction From Genome Data: Contrasting EBV in Livestock to PRS in Humans: Genomic Prediction.” Genetics 211: 1131–1141. [DOI] [PMC free article] [PubMed] [Google Scholar]
  99. Yokomori, T. , Ohnuma A., Tozaki T., Segawa T., and Itou T.. 2023. “Identification of Personality‐Related Candidate Genes in Thoroughbred Racehorses Using a Bioinformatics‐Based Approach Involving Functionally Annotated Human Genes.” Animals (Basel) 13: 769. [DOI] [PMC free article] [PubMed] [Google Scholar]
  100. Zhou, Z. , Jin Z., Tian Y., Huangfu C., Fan Z., and Liu D.. 2024. “CDK14 Is Regulated by IGF2BP2 and Involved in Osteogenic Differentiation via Wnt/Beta‐Catenin Signaling Pathway In Vitro.” Life Sciences 358: 123148. [DOI] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

Table S1: Details for the discovery, training and validation sets.

Table S2: Allelic association test results for the Discovery set. The top SNP (lowest p‐value) in each ROI is shown. SNPs are shown ranked by p‐value for association with discipline.

Table S3: Candidate genes for jumping. Candidate genes identified based on proximity to the lead SNP, and biological function relevant to jumping. Highly significant SNPs are in bold.

Table S4: Candidate genes for dressage. Candidate genes identified based on proximity to the lead SNP, and biological function relevant to dressage. Highly significant SNPs are in bold.

Table S5: Summary statistics for PPS models using different SNP sets for a range of p‐value cut‐offs. The p‐value shown is the result of a t‐test comparing the PPS between Jump and Dressage horses in the Training set.

Table S6: Confusion matrix showing how many horses were correctly and incorrectly assigned to each discipline using the median PPS as the cut‐off value.

Table S7: Accuracy to predict Jump and Dressage using the median PPS as the cut‐off value.

Figure S1: Principal component analysis (PCA) plot illustrating the distribution of the genetic variation among Jump and Dressage horses for the first two PCs that accounted for 28.6% of the genetic variance.

Figure S2: Proportion of Jump and Dressage horses within each prediction category.

Figure S3: Proportion of horses with jumping competition records categorised as Jump from the PPS.

Figure S4: Genetic predictions of second careers for Thoroughbreds in Australia, Europe and USA, using the PPS developed for sport horses (D Pro—dark blue, D—orange, J—green, J Pro—light blue).

AGE-57-0-s001.docx (155KB, docx)

Data Availability Statement

The data that support the findings of this study are not publicly available due to privacy restrictions. The data may be made available for academic purposes after signing a Material Transfer Agreement, which will be subject to confidentiality and will prevent third parties from commercial exploitation.

Articles from Animal Genetics are provided here courtesy of Wiley

RESOURCES