https://orcid.org/0000-0002-5158-4558
Figures
Abstract
The Korean Sapsaree dog is a native breed known for its distinctive appearance and historical significance in Korean culture. The accurate estimation of breeding values is essential for the genetic improvement and conservation of such indigenous breeds. This study aimed to evaluate the accuracy of breeding values for body height, body length, chest width, hair length, and distraction index (DI) traits in Korean Sapsaree dogs. Additionally, a genome-wide association study (GWAS) was conducted to identify the genomic regions and nearby candidate genes influencing these traits. Phenotypic data were collected from 378 Korean Sapsaree dogs, and of these, 234 individuals were genotyped using the 170k Illumina CanineHD BeadChip. The accuracy of genomic predictions was evaluated using the traditional BLUP method with phenotypes only on genotyped animals (PBLUP-G), another traditional BLUP method using a pedigree-based relationship matrix (PBLUP) for all individuals, a GBLUP method based on a genomic relationship matrix, and a single-step GBLUP (ssGBLUP) method. Heritability estimates for body height, body length, chest width, hair length, and DI were 0.45, 0.39, 0.32, 0.55, and 0.50, respectively. Accuracy values varied across methods, with ranges of 0.22 to 0.31 for PBLUP-G, 0.30 to 0.57 for PBLUP, 0.31 to 0.54 for GBLUP, and 0.39 to 0.67 for ssGBLUP. Through GWAS, 194 genome-wide significant SNPs associated with studied Sapsaree traits were identified. The selection of the most promising candidate genes was based on gene ontology (GO) terms and functions previously identified to influence traits. Notable genes included CCKAR and DCAF16 for body height, PDZRN3 and CNTN1 for body length, TRIM63, KDELR2, and SUPT3H for chest width, RSPO2, EIF3E, PKHD1L1, TRPS1, and EXT1 for hair length, and DDHD1, BMP4, SEMA3C, and FOXP1 for the DI. These findings suggest that significant QTL, combined with functional candidate genes, can be leveraged to improve the genetic quality of the Sapsaree population. This study provides a foundation for more effective breeding strategies aimed at preserving and enhancing the unique traits of this Korean dog breed.
Citation: Haque MA, Kim N-K, Yeji R, Lee B, Ha J-H, Lee Y-M, et al. (2024) Genomic prediction and genome-wide association studies of morphological traits and distraction index in Korean Sapsaree dogs. PLoS ONE 19(11): e0312583. https://doi.org/10.1371/journal.pone.0312583
Editor: Haitham Abo-Al-Ela, Suez University Faculty of Fish Resources, EGYPT
Received: May 16, 2024; Accepted: October 10, 2024; Published: November 21, 2024
Copyright: © 2024 Haque et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Data Availability: The data used in this study are accessible on Figshare (https://doi.org/10.6084/m9.figshare.25769994.v2).
Funding: This study was financially supported by Yeungnam University (https://www.yu.ac.kr) in the form of a research grant (2021) received by J-JK. No additional external funding was received for this study.
Competing interests: The authors have declared that no competing interests exist.
Introduction
The domestic dog (Canis lupus familiaris) has always been considered a human’s best friend since their domestication due to their loyalty to humans. In the animal kingdom, dogs were the earliest animals to be domesticated that have been dated to about 14,000 years ago [1, 2]. Corroboration from texts, artifacts, and paintings from the tomb disclose that people care for dogs as loved pets and consider them as members of the beloved family and involve them in various family events. In contemporary civilization, dogs play significant roles in human needs for companionship, affection, friendship, and love, which have become increasingly difficult to say in “our nuclear families living impersonal suburban lifestyles” [3]. As of today, more than four hundred (400) recorded dog breeds around the globe have been bred for specialized purposes including herding, rescuing, guarding, hunting, pulling sleds, retrieving, and military activities [4].
The Korean peninsula has some of the most familiar dog breeds in the world, with approximately more than 150 breeds [5]. Among them, the Sapsaree, Jindo, and Donggyeong are designated as ‘natural monuments’ by the Korean government (Cultural Heritage Administration of Korea, #368, 54 and 540, respectively) [6, 7]. The Korean Kennel Federation established in 1945 under the Ministry of Agriculture & Forestry is the representative canine organization of South Korea and is responsible for the preservation and promotion of native Korean dogs. The Sapsaree is one of the familiar indigenous dog breeds in the Korean peninsula. These droopy-eared, shaggy-haired Sapsaree dogs are presumed to reflect the personality of Korean persons [5]. Generally, the dogs of this breed are human-friendly and also appear to be loyal, gentle, and protective and they develop strong bonds with their master. They are medium-sized dogs with mature males’ heights between 20 and 23 inches, body weights between 40 and 62 pounds, and females’ heights up to 22 inches and weights between 35 to 55 pounds [5].
Initially, this breed was kept only by royals and aristocrats, but nowadays the Sapsaree has become a family member and widespread household pet for all social levels within Korea. During the Japanese colonial period (from 1910 to 1945) and the Korean War (from 1950 to 1953), the Korean Sapsaree population was close to disappearance because most long, soft, and shaggy-haired dogs were killed for their hides and fur from their coats for making winter clothes for the Japanese military. When it reached the verge of extinction, a member of the Sapsaree breed could rarely be found in 1986 by the regional Sapsaree lovers in the Daegu area, Korea.
Understanding the genetic basis of canine traits such as body height, body length, chest width, hair length, and distraction index is crucial for improving breeding practices, enhancing welfare, and addressing breed-specific health concerns. These traits play a pivotal role in determining the overall conformation, functionality, and aesthetic appeal of dogs across various breeds. In particular, recent studies have highlighted the role of specific genetic markers in determining desirable traits, paving the way for more effective breeding strategies that can address both performance and health issues [8]. Genome-wide association studies (GWAS) and genomic prediction have emerged as indispensable tools in elucidating the genetic architecture underlying these traits, thereby facilitating informed breeding decisions and the selection of desirable phenotypes within dog populations. Identifying genetic markers associated with body height, body length, chest width, and hair length allows breeders to optimize breeding programs to enhance desirable traits while minimizing the risk of inherited disorders. For instance, understanding the genetic determinants of body conformation can aid in the development of breeding strategies to improve structural soundness [9] and reduce the prevalence of musculoskeletal issues such as hip dysplasia. Similarly, unraveling the genetic basis of hair length can inform breeding decisions to produce dogs with coat types that are both aesthetically pleasing and well-suited to their environmental conditions.
Furthermore, the distraction index–a measure of hip laxity obtained through radiographic assessment–is of paramount importance in the prevention and management of canine hip dysplasia (CHD). Canine hip dysplasia (CHD) or distraction index (DI) was initially reported in 1935 by Dr. Gerry B. Schnelle [10] is a common inborn skeletal condition seen across all purebred and mixed-breed dogs designated by irregular development of the hip joint and joint laxity. When normally developing the hip structures, several dogs face an innate looseness or spillover laxity in the joint causes irregular joint forces, and beginning the body size and weight increase, the repetitive microdamage to joint structures and the cartilaginous matrix of the hip are formed in misshaped [11]. Gradually this abnormal structure causes severe osteoarthritis [12, 13]. This hip disease is a major problem in many dog breeds, especially in medium and large dog breeds. By identifying genetic variants associated with hip laxity, GWAS and genomic prediction can assist breeders in selecting breeding pairs with lower predispositions to CHD, thereby reducing the incidence of this debilitating condition in future generations of dogs. Additionally, early detection of hip laxity through genetic screening can facilitate proactive management strategies, such as lifestyle modifications and surgical interventions, to mitigate the progression of CHD and improve the quality of life for affected dogs [14].
Moreover, advancements in genomic technologies have enabled more precise identification of SNPs that correlate with specific traits, enhancing the accuracy of genomic predictions [15]. Recent advances in Sapsaree dog research have predominantly focused on genetic diversity and population structure [5], as well as the estimation of effective population size [16]. However, the application of GWAS and genomic prediction to the Sapsaree breed is still in its infancy, with significant gaps in understanding the specific traits that are genetically determined and their heritability. The exploration of these traits could provide valuable insights into breed management and conservation efforts. The accuracy of genomic prediction and GWAS concerning traits such as body height, body length, chest width, hair length, and DI remains largely unexplored. The scarcity of literature addressing the accuracy of genomic predictions and GWAS for these traits prompted the primary objective of our investigation. Therefore, our objectives were to estimate the heritability, and genomic prediction accuracy and identify significant SNPs associated with these traits. Additionally, we aimed to explore the genetic architecture and biological relevance of these markers at the whole-genome level and identify potential candidate genes in the Sapsaree population. These findings will facilitate genetic improvement and selective breeding strategies, leading to enhanced productivity and performance within this breed.
Materials and methods
Ethics statement
The care and management of all animals used in this study were approved by the Animal Care and Use Committee of the National Institute of Animal Science (NIAS), Rural Development Administrations (RDA), South Korea (Approval No. 2016–177). Ear tissue samples were collected by veterinarians following ethical guidelines for animal health and welfare. The experimental animals were not anesthetized or euthanized in this study. We confirmed that all methods are reported in accordance with ARRIVE guidelines (https://arriveguidelines.org) for the reporting of animal experiments.
Animals and phenotypes
Phenotypic data were collected from 378 Korean Sapsaree dogs from the Korean Sapsaree Foundation. Pedigree data from 872 individuals were used in the animal model. The phenotypic data included measurements for body height, body length, chest width, hair length, and DI traits. Body height, body length, and chest width were measured in centimeters (cm), while hair length traits were categorized as short (n = 224) or long hair (n = 154). The study employed the PenHIP (University of Pennsylvania Hip Improvement Program) radiographic method to measure CHD values ranging from 0 to 1. Lower values indicate tighter hip joints and reduced CHD risk.
Animal pedigree
The pedigree of the studied animals was obtained from the Korean Sapsaree Foundation. This pedigree, which included both phenotypic and genotypic data, covered a total of 1,088 animals and extended up to a maximum of 12 ancestral generations. Within this dataset, 429 animals were found inbred. Additionally, the dataset included 313 sires, 388 dams, and 166 full-sib family groups, with an average family size of 3.25. Although the highest observed inbreeding coefficient in the study was 0.25, the average coefficient for the entire population was 0.02, while within the inbred subset, it was slightly higher at 0.05. Furthermore, the average relatedness among all animals was calculated based on the pedigree data. The pedigree data was processed using the PEDIG software [17], while the pedigree structure was calculated using CFC software [18].
Genotypes
A total of 234 Korean Sapsaree dogs were genotyped using an Illumina 170K CanineHD BeadChip (Illumina Inc., San Diego, CA, USA), containing 173,662 embedded SNPs. To ensure data quality, we initially removed SNPs located on sex chromosomes in duplicate or uncertain positions, resulting in the elimination of 6,478 SNPs. This left us with 167,184 SNPs for analysis. Subsequently, multiple quality control (QC) criteria were implemented to filter out low-quality SNPs. Specifically, SNPs with a minor allele frequency (MAF) of less than 5% (i.e., monomorphic), a SNP call rate below 90%, individuals with a genotyping call rate less than 90%, and SNPs showing a significant deviation from Hardy–Weinberg Equilibrium (HWE) with a p-value greater than 10−6 were excluded from the dataset. Additionally, the identity-by-state (IBS) test was conducted to identify genetically similar individuals or genotyping errors. The pair of individuals showing a similarity rate >99% indicated either an identical animal or an error in genotyping. The IBS and entire QC process were performed using the PLINK v1.9 toolset [19]. Subsequently, all genotyped animals were imputed using Beagle v5.1 software [20]. After IBS and QC procedures, a total of 205 animals with genotypes for 104,003 SNPs were available for further analysis.
Statistical analysis
Estimation of variance components.
The variance components and heritabilities for body height, body length, chest width, and DI traits were estimated using the restricted maximum likelihood method (REML) for animal models using BLUPF90+ v2.52 software [21]. For hair length traits, which are categorical, the data were converted to binary format, assigning a value of 1 for long hair and 0 for short hair. The probability of the binary dataset was modeled using a generalized linear mixed model (GLMM) with a logit function, followed by fitting an animal model for variance components in ASReml-SA v4.2 [22]. Since the trait records were converted to a 0–1 scale, incidence frequency was not included in the analysis; instead, values of 0 and 1 were used. A single-trait pedigree-based animal model is as follows:
(1)
where, y represents the vector of phenotypes; b represents the vector of fixed effects including the sex of the animals, measurement date, and age as a covariate; u represents the vector of additive genetic effects of the individuals; X is the incidence matrix of b; Z is the incidence matrix of u; and e is the vector of the residuals. It was assumed that
and
, where A is the pedigree-based genetic relationship matrix and
is the additive genetic variance, and
is the residual variance. The adjusted phenotypes yc were obtained for each trait and animal as the residual effects of the y = Xb + e, where
[23].
The heritability for body height, body length, chest width, and DI traits were calculated using the following formula [24]:
where,
is the additive genetic variance;
is the residual variance; and
is the phenotypic variance.
The heritability for hair length (0 & 1) was calculated by the following equation [25]:
where,
is the additive genetic variance; Φp is the incidence in the population; and Φp (1 − Φp) is the phenotypic variance on the observed scale.
Estimation of breeding values.
We employed four methods to estimate breeding values: a traditional BLUP method using phenotypes only on genotyped animals (PBLUP-G), another traditional BLUP method utilizing a pedigree-based relationship matrix (PBLUP) of all individuals, a GBLUP method based on a genomic relationship matrix, and a single-step GBLUP (ssGBLUP) method. The ssGBLUP method combines the relationship matrix constructed from genotyped and non-genotyped individuals with pedigree information to predict breeding values.
The BLUP model used to predict conventional EBV was as follows [23]:
(2)
where, yc represents the vector of trait observations adjusted for fixed effects, 1 is the vector of ones; μ is the overall mean, and the remaining notations are consistent with those in Eq 1. Additionally, EBVs were obtained solely using phenotypic data and pedigree information for genotyped animals (PBLUP-G).
To estimate genomic breeding values, we used the model (2) above under the following assumptions:
u represents the vector of additive genetic effects solely for genotyped individuals, and Z denotes the incidence matrix of u. It was assumed that where G was the genomic relationship matrix (GRM) constructed using SNP information as follows [26]:
(3)
where n is the total number of markers (104,003); pi is the allele frequency of the ith marker; and M is the matrix of centered genotypes.
In the ssGBLUP method, the statistical model resembled that used for traditional evaluation. However, both genotyped and non-genotyped animals were simultaneously incorporated into the hybrid relationship matrix H, which was a combination of the numerator relationship matrix A and the genomic relationship matrix G. The inverse of the H matrix was derived using the following equation [27] and employing the preGSf90 software [28]:
(4)
where A22 is the numerator relationship matrix of genotyped animals.
Accuracy of prediction
We implemented a repeated fivefold cross-validation (CV) approach to assess genomic prediction accuracy. In this procedure, the entire population was randomly divided into five equal groups for the five-fold cross-validation. Thus, one portion of the data (20%) served as the validation or testing group, while the remaining four portions (80%) constituted the reference or training group [9]. This process was repeated five times to ensure that each animal in the dataset had an opportunity to be included in both the testing and reference groups. The prediction accuracy was assessed using Pearson’s correlation coefficient (r) between the adjusted phenotypes of individuals in the validation dataset and their (G)EBV divided by the square root of the heritability for each trait. The empirical standard error (SE) was determined by dividing the standard deviation of the five calculated accuracies from the fivefold cross-validation (CV) by the square root of 5. Additionally, the slope of the regression of phenotype on (G)EBV was computed to measure the bias in the (G)EBV. A regression coefficient close to 1 indicates no bias, while a slope of <1 or >1 suggests the underestimation or overestimation of (G)EBV, respectively.
GWAS analysis
The Sapsaree traits underwent analysis using the linear mixed model (LMM) implemented in the genome-wide efficient mixed-model analysis (GEMMA) software v0.98.5 [29]. GEMMA calculates a genomic relationship matrix (GRM) between individuals within each population to determine the population structure. The univariate linear mixed model in GEMMA was described as follows:
(5)
where y is the vector of phenotypes; W is the vector of the fixed effects of sex of the animals, measurement date, and age as a covariate; α is a vector of the corresponding coefficients, including the intercept; X represents the vector of all marker genotypes; β represents the effect size of the SNP; u ~ MVNn (0, λτ−1 K) is an n-vector of animal additive effects; and ε ~ MVNn (0, τ−1 In) represents an n-vector of errors; τ−1 is the variance of the residual errors; λ is the ratio between the two variance components; K represents the genomic relationship matrix (GRM); In is an n ⅹ n identity matrix; and MVNn represents the n-dimensional multivariate normal distribution. GEMMA calculates the GRM as follows [29]:
(6)
where X represents the n × p matrix of the genotypes, xi represents the genotypes of the ith SNP,
is the sample mean, and 1n is the n × 1 vector of 1.
Identification of candidate genes and analysis of functional enrichment
We identified putative candidate genes within the QTL regions and in the nearest upstream and downstream regions (500 kb) of our mapped significant SNPs [30]. This analysis utilized the dog genome (Canfam3.1) assembly and employed online resources such as the NCBI Genome Data Viewer (https://www.ncbi.nlm.nih.gov/genome/gdv?org=canis-lupus-familiaris; accessed on 18 March 2024). Manhattan plots were generated to visualize the genome-wide distribution of significant SNPs, with the significance level represented as the negative base -10 logarithm (-log10) of the p-value for each SNP. Additionally, quantile-quantile (QQ) plots were generated to illustrate the observed versus expected p-values (-log10P) for each trait. Furthermore, we calculated the genomic inflation factor, lambda (λ), to evaluate population stratification by comparing the median chi-squared test statistics from GWAS to the expected median of the chi-squared distribution. The p-values from GWAS results for all traits were utilized to compute λ using the qchisq() function in R [31]. Subsequently, we conducted KEGG and GO analyses using DAVID [32] and KOBAS v3.0 [33] to explore the functions of all candidate genes. Enriched terms with a significance threshold of p-value < 0.05 were selected to further explore the genes involved in pathways and biological processes. The functional roles of the identified genes within and near significant SNPs associated with reproductive traits were explored using published reports in PMC for Biotechnology Information (NCBI database) journals and other literature surveys. The functional roles of each gene were also obtained from online resources, including human gene functions at GeneCards (www.genecards.org), the Mouse Genome Informatics (MGI) website (https://www.informatics.jax.org/), and Ensembl (www.ensembl.org/biomart/martview), accessed on 19 March 2024. Candidate genes showing functional relevance to the desired traits of interest were considered promising candidate genes.
Results
Descriptive statistics
Table 1 represents the summary statistics for the studied traits to estimate the variance components and the estimation of breeding values. The pedigree-based analysis method yielded mean body height, body length, chest width, and DI values of approximately 55.18 cm, 61.83 cm, 21.39 cm, and 0.50, respectively. The standard deviations for these traits ranged from 0.11 to 3.99, suggesting moderate variability within the population. Meanwhile, the genome-based analysis method produced slightly different mean values for body height (55.17 cm), body length (61.57 cm), chest width (21.19 cm), and DI (0.49). Standard deviations for these traits were also comparable to the pedigree-based method, ranging from 2.84 to 3.88 cm for body measurements and 0.11 for DI. Additionally, the pedigree-based analysis revealed that there were 224 long-haired dogs, constituting about 59.3% of the sample, while approximately 154 were short-haired dogs, representing around 40.7%. Conversely, the genome-based analysis method, which included 205 individuals, indicated that approximately 82 were short-haired dogs, accounting for roughly 40% of the population, and approximately 123 were long-haired dogs, making up about 60%. These statistics offer insights into the morphological characteristics of the Sapsaree dog population under each analytical approach, highlighting consistency in trait measurements and variability across methods.
Estimation of heritability
The heritability estimates provide valuable insights into the genetic basis of various traits observed within the Sapsaree dog population. The pedigree-based heritability assessments revealed diverse values across the studied traits, indicating varying degrees of influence from genetic factors. These estimates, ranging from 0.32 to 0.55 (Table 2), reflect the proportion of trait variation that can be attributed to genetic differences among individuals within the population. A higher heritability estimate, such as that of hair length (0.55), indicates a stronger genetic influence on the trait. In the case of body height (0.45), for example, the estimate suggests that genetic factors play a significant role in determining the height of Sapsaree dogs, while traits such as body length (0.39) and chest width (0.32) exhibit slightly lower heritability estimates, indicating a moderate genetic influence. Similarly, the DI, a measure of hip joint laxity, also shows a notable genetic component (0.50), indicating that genetic factors contribute significantly to individual differences in this aspect of hip morphology within the Sapsaree population. These findings highlight the complex interplay between genetics and phenotype and underscore the importance of genetic factors in shaping the observed traits in Sapsaree dogs. Understanding the genetic basis of these traits is crucial for informed breeding decisions aimed at maintaining or enhancing desirable characteristics within the population.
Evaluation of genomic prediction accuracy
The predictive accuracies for the studied traits obtained using PBLUP-G, PBLUP, GBLUP, and ssGBLUP methods are shown in Fig 1. Regarding the body height trait, the ssGBLUP model demonstrates the highest prediction accuracy of 0.60, closely followed by PBLUP with an accuracy of 0.53. PBLUP-G and GBLUP also exhibit reasonable accuracies of 0.28 and 0.51, respectively. Similarly, for body length, ssGBLUP leads with the highest accuracy of 0.55, followed closely by PBLUP at 0.46. PBLUP-G and GBLUP trail behind with accuracies of 0.23 and 0.44, respectively. In the case of chest width, all models show relatively lower accuracies compared to other traits, with ssGBLUP achieving an accuracy of 0.39. PBLUP-G and PBLUP follow with accuracies of 0.22 and 0.30, respectively, while GBLUP exhibits a slightly improved accuracy of 0.31. Conversely, hair length demonstrates a notable improvement in prediction accuracy, with ssGBLUP achieving an accuracy of 0.67. PBLUP also shows a strong performance with an accuracy of 0.57, which is roughly equivalent to that of GBLUP at 0.59. Meanwhile, PBLUP-G shows a lower accuracy of 0.31. Additionally, for the DI trait, ssGBLUP stands out as the top performer with an accuracy of 0.63, followed closely by PBLUP at 0.54 and GBLUP at an intermediate accuracy of 0.57, while PBLUP-G lags behind with an accuracy of 0.27.
Vertical lines indicate the empirical standard error for cross-validation results.
Bias in genomic prediction accuracy
The bias in genomic predictions for all studied traits is shown in Table 3. Noteworthy variations in biases among different models for each trait are apparent upon comparison. It’s important to note that prediction bias plays a pivotal role in determining the use of estimated breeding values for genetic evaluation. To measure prediction bias, we calculated the regression coefficient of adjusted phenotypes. This coefficient serves as a reliable measure of the degree of bias inherent in the predictions made by each model. Across all studied traits, the regression coefficients of prediction varied depending on the model employed. Specifically, the regression coefficients of prediction ranged from 0.65 to 0.94 for PBLUP-G, 0.49 to 0.68 for PBLUP, 0.85 to 0.99 for GBLUP, and 0.79 to 0.98 for ssGBLUP methods. On average, the regression coefficients were 0.79, 0.72, 0.85, and 0.80 using PBLUP-G, PBLUP, GBLUP, and ssGBLUP methods, respectively.
GWAS results
For the GWAS analysis of the traits studied in the Sapsaree population, we identified the genome-wide distribution of significant SNPs using the Bonferroni test to determine the genome-wide significance threshold (0.05/N, where N represents the number of SNPs). With the exception of the hair length trait, there were only a few significant SNPs passing the Bonferroni test for the other studied traits. Notably, neither the chest width nor the DI trait yielded any significant SNPs meeting the Bonferroni test criteria. Hence, we employed a more lenient threshold at the suggestive level (1/N, where N represents the number of SNPs) to identify suggestive SNPs, recognizing the strictness of the Bonferroni criterion [34]. Although the DI failed to meet the suggestive level threshold and the chest width yielded only a single significant SNP, we chose a significant threshold level of -log10P value of 4 for these traits. While this suggestive significance threshold may potentially lead to false positive results, we aimed to encompass all plausible loci underlying the studied traits in Sapsaree, hence adopting a less stringent threshold to minimize the risk of missing potential trait-associated markers through GWAS [35]. These results are further described as follows.
Marker loci associated with body height.
The GWAS conducted for body height in the Sapsaree identified a total of 28 significant SNPs at the suggestive level across the genome. Notably, CFA3 exhibited the highest number of SNPs, with 25 variants detected, suggesting a potential genomic hotspot for genetic factors influencing body height. These genetic markers spanned a genomic range, with positions ranging from 82.05 Mb to 91.47 Mb. Additionally, significant SNPs were observed at 8.62 Mb on CFA23 and from 13.53 Mb to 13.89 Mb on CFA23. Among the 28 significant SNPs identified, three marker loci surpassed the Bonferroni significant threshold. These markers include BICF2P540899 (p = 2.96 × 10−7), BICF2P1218152 (p = 2.96 × 10−7), and BICF2G630353913 (p = 3.11 × 10−7), located within the genomic region of 84.04 Mb to 84.59 Mb (Table 4, Fig 2a). The genomic inflation factor (λ) for body height was calculated to be 0.981 (Fig 2b). To visually represent the observed versus expected p-values (-log10P) for body height, QQ plot was generated (Fig 2b). The QQ plot clearly demonstrated a close alignment between the observed and expected values, indicating a normal distribution of p-values. Additionally, the upper tails of the plot show an increase in significance, which is consistent with expectations for a GWAS. This alignment suggests that population stratification was effectively addressed using the appropriate model, thereby enhancing the likelihood of identifying true associations.
(a) Manhattan plot. The y-axis represents -log10 (observed) p-values for genome-wide SNPs against their respective positions on each chromosome (x-axis). The horizontal dark green line indicated the suggestive (p = 9.615 × 10−6) threshold level and the yellow green line indicated the Bonferroni corrected (p = 4.8 × 10−7) threshold level. (b) Quantile–quantile plot. The red line represents the 95% concentration band under the null hypothesis of no association. The blue dot represents the p-values of the entire study.
Marker loci associated with body length.
We identified six significant SNPs at the suggestive level across three different CFAs, with a dense cluster observed on CFA27 (4). Additionally, single significant marker loci associated with body length in Sapsaree were detected on CFA20 and CFA23 (Table 5). While we initially established a suggestive threshold limit to identify genome-wide marker loci influencing body length in Sapsaree, we also performed a Bonferroni correction. Notably, three significant loci surpassed the Bonferroni corrected significant level, situated on CFA20 (1) and CFA27 (2). These markers include BICF2P348989 (p = 1.04 × 10−8), BICF2G630142608 (p = 7.54 × 10−9), and BICF2P5196 (p = 1.32 × 10−8) (Fig 3a). The body length exhibited a λ value of 0.996 (Fig 3b). The QQ plot representing the observed versus expected p-values (-log10P) for body length (Fig 3b) suggests that population stratification was effectively addressed using the appropriate model. Additionally, the upper tails of the plot demonstrate an increase in significance, as expected for GWAS, thereby enhancing the likelihood of identifying true associations.
(a) Manhattan plot. The y-axis represents -log10 (observed) p-values for genome-wide SNPs against their respective positions on each chromosome (x-axis). The horizontal dark green line indicated the suggestive (p = 9.615 × 10−6) threshold level and the yellow green line indicated the Bonferroni corrected (p = 4.8 × 10−7) threshold level. (b) Quantile–quantile plot. The red line represents the 95% concentration band under the null hypothesis of no association. The blue dot represents the p-values of the entire study.
Marker loci associated with chest width.
A total of 29 significant SNPs, each with a -log10P-value of 4 were detected for chest width in the Sapsaree population. Interestingly, within the scope of this investigation, none of the SNPs met the criteria for the Bonferroni correction threshold, with only a single SNP meeting the suggestive significance threshold. A dense cluster of significant SNPs was observed primarily on CFA31 (9), followed by CFA18 (5), CFA6 (4), and CFA2, CFA12, and CFA37, each with 3 significant marker loci associated with chest width (Table 6). Additionally, CFA22 harbored only two significant marker loci linked to chest width. Although, we set a -log10P value of 4 as the limit for identifying genome-wide marker loci affecting chest width in Sapsaree, we also calculated a suggestive significance level. Notably, on CFA18, the SNP BICF2G630697745 (p = 8.76 × 10−6) exhibited suggestive significance (Fig 4a). We intended to report loci with a less stringent threshold to capture most of the genetic variation and understand the genetic architecture of chest width in Sapsaree, thereby reducing the risk of overlooking any potential markers during analysis. Furthermore, the chest width exhibited a λ value of 1.008, indicating minimal inflation due to population stratification. The QQ plot depicted the observed versus expected p-values (-log10P), as shown in Fig 4b.
(a) Manhattan plot. The y-axis represents -log10 (observed) p-values for genome-wide SNPs against their respective positions on each chromosome (x-axis). The horizontal dark green line indicated the considered (p = 9.90 × 10−5) threshold level and the yellow green line indicated the suggestive (p = 9.615 × 10−6) threshold level. (b) Quantile–quantile plot. The red line represents the 95% concentration band under the null hypothesis of no association. The blue dot represents the p-values of the entire study.
Marker loci associated with hair length.
For hair length, we identified a total of 96 significant SNPs that surpassed the Bonferroni correction threshold and were situated on CFA13 (Fig 5a). Given the ample number of SNPs detected above the Bonferroni correction threshold, we did not consider the suggestive significance level to identify significant SNPs associated with hair length. Notably, the most significant SNPs (BICF2G630607427, BICF2G630607436; p = 3.17 × 10−17) were located on CFA13 within the genomic region spanning from 11.57 Mb to 11.59 Mb position (Table 7). Our findings highlight CFA13 as a region where highly significant SNPs are clustered, suggesting a potential genomic hotspot for genetic factors influencing hair length in the Sapsaree population. Additionally, we calculated λ, which yielded a value of 0.920 (Fig 5b). The QQ plot illustrated the observed versus expected p-values (-log10P), as depicted in Fig 5b.
(a) Manhattan plot. The y-axis represents -log10 (observed) p-values for genome-wide SNPs against their respective positions on each chromosome (x-axis). The horizontal dark green line indicated the suggestive (p = 9.615 × 10−6) threshold level and the yellow green line indicated the Bonferroni corrected (p = 4.8 × 10−7) threshold level. (b) Quantile–quantile plot. The red line represents the 95% concentration band under the null hypothesis of no association. The blue dot represents the p-values of the entire study.
Marker loci associated with DI.
We selected SNPs with a p-value (-log10P) of 4.0 and higher as significant markers associated with the DI trait. This less stringent threshold was chosen to encompass most of the genetic variation and gain insights into the genetic architecture of DI in Sapsaree, thereby minimizing the risk of overlooking potential markers during analysis. The GWAS analysis for the DI trait identified 25 significant marker loci distributed across four CFAs. Notably, CFA8 harbored the most significant loci (8), followed by CFA18 and CFA20 with 6 each, while CFA7 identified 5 SNPs contributing to a total of 35 significant SNPs associated with DI in Sapsaree (Fig 6a). Among the significant SNPs identified on CFA8, they were clustered within a genomic region spanning from 26.79 Mb to 31.19 Mb (Table 8). Additionally, the DI trait exhibited a λ value of 0.989, indicating minimal inflation due to population stratification. The QQ plot illustrated the observed versus expected p-values (-log10P), as depicted in Fig 6b.
(a) Manhattan plot. The y-axis represents -log10 (observed) p-values for genome-wide SNPs against their respective positions on each chromosome (x-axis). The horizontal dark green line indicated the considered (p = 9.90 × 10−5) threshold level and the yellow green line indicated the suggestive (p = 9.615 × 10−6) threshold level. (b) Quantile–quantile plot. The red line represents the 95% concentration band under the null hypothesis of no association. The blue dot represents the p-values of the entire study.
Candidate genes and functional annotation
In our study, we investigated candidate genes situated near regions of association, as these genes can significantly influence the expression of complex phenotypes due to their known biological and physiological properties. To identify these genes, we conducted a search in the NCBI database on dogs (Canis lupus familiaris) using the CanFam3.1 genome assembly. We focused on a search window of ±500 Kb around the identified marker loci associated with the studied traits in the Sapsaree population. This approach allowed us to comprehensively explore the genomic landscape surrounding the significant marker loci and gain insights into potential candidate genes influencing the traits under investigation.
We identified a total of 177 unique positional candidate genes within a 1 Mb region centered in proximity to the significant marker loci in Sapsaree (Tables 4–8). Specifically, we identified 20 nearby genes associated with body height, 5 genes with body length, 28 genes with chest width, 103 genes with hair length, and 21 genes with DI. Prominent genes associated with body height comprised CCKAR and DCAF16, while PDZRN3 and CNTN1 were notable for body length. For chest width, significant genes included TRIM63, KDELR2, and SUPT3H. Hair length was associated with RSPO2, EIF3E, PKHD1L1, TRPS1, and EXT1, while the DI trait was linked to DDHD1, BMP4, SEMA3C, and FOXP1 (Table 9). These identified genes play significant roles in shaping the phenotype of Sapsaree dogs and provide valuable insights into the genetic mechanisms underlying these traits.
Functional annotation of Gene Ontology (GO) terms was initially conducted to discern the biological significance and systematic features of the candidate genes. This was accomplished using the Database for Annotation, Visualization, and Integrated Discovery (DAVID) and the KEGG Orthology-Based Annotation System (KOBAS). The gene ontology comprised four categories: Biological Process (BP), Molecular Functions (MF), Cellular Component (CC), and Kyoto Encyclopedia of Genes and Genomes (KEGG). Candidate genes associated with reproductive traits exhibited significantly distinct GO terms (P < 0.05). Specifically, 66 GO terms were associated with Biological Process (BP), 7 with Cellular Component (CC), 12 with Molecular Functions (MF), and 4 with KEGG pathways (S1 Table). It is noteworthy that the majority of common genes identified for the studied traits play crucial roles in development and growth-related processes, including heart development, embryonic development, tissue development, epithelial development, limb development, skeletal system development, nervous system development, and organ development in animals.
Discussion
Based on the findings of this study, it is evident that the studied traits demonstrated varying degrees of heritability, ranging from moderate to high, which aligns with findings from prior research focusing on similar traits. For instance, Verryn and Geerthsen [36] reported a notably high heritability of 0.81 for chest width in German Shepherd dogs, mirroring the robust genetic influence detected in our investigation. Likewise, Famula [37] found moderate to high heritability estimates ranging from 0.35 to 0.65 for various body traits in Labrador Retrievers, demonstrating consistency with our findings across different breeds. However, a significant discrepancy emerged in the heritability estimation for the DI trait compared to previous studies. Specifically, Ginja et al. [38] reported a remarkably high heritability of 0.83 ± 0.11 for the DI trait in Estrela Mountain Dogs, contrasting sharply with Todhunter et al’s. [39] finding of heritability of 0.50 Labrador Retrievers, which closely resembles our estimation for Sapsaree dogs. This disparity underscores the complexity of genetic factors influencing hip morphology across different dog breeds.
Similarly, Zhang et al. [40] reported a heritability of 0.61 for the DI trait in dogs admitted to Cornell University Hospital using a multiple-trait model. Leighton et al. [41] employed a Bayesian two-trait model and reported heritability estimates of 0.60 for German Shepherds, 0.66 for Labrador Retrievers, and 0.59 for Golden Retrievers for the DI trait. Tikekar et al. [42] estimated the heritability of the DI trait in German Shepherds using a linear mixed model, obtaining values of 0.21 ± 0.32 for the right hip and 0.67 ± 0.38 for the left hip. Additionally, Tikekar et al. [43] reported a wide range of heritability estimates for the DI trait in German Shepherd dogs in New Zealand, ranging from 0.15 to 0.81. Wierzbicki [44] reported that hair length heritability in Polish fur animals revealed a comparatively lower heritability of 0.337, emphasizing the nuanced genetic architectures governing phenotypic traits across distinct species and populations.
The observed variations in heritability estimates can be attributed to a multitude of factors, including but not limited to population structure, breed-specific characteristics, environmental influences, sample size disparities, measurement inaccuracies, and methodological approaches utilized in data analysis [23]. Ruefenacht et al. [45] emphasized the critical importance of achieving consistency in heritability estimates across different evaluation methodologies to bolster the reliability of genetic parameters in breeding programs and population management strategies. Moving forward, it is imperative for future studies to prioritize larger sample sizes drawn from the Sapsaree population to enhance the precision and robustness of heritability estimations, thus facilitating more informed decision-making in breeding practices and conservation efforts.
In this research, we conducted an investigation into the accuracy of breeding values associated with various traits including body height, body length, chest width, hair length, and DI. Four distinct models were employed for this investigation, including pedigree-based BLUP solely on genotyped animals (PBLUP-G), pedigree-based BLUP incorporating all animals (PBLUP), genomic BLUP (GBLUP), and single-step genomic BLUP (ssGBLUP). Subsequently, the accuracies of the breeding values derived from these models were meticulously evaluated and compared. Guo et al. [46] previously reported a range of prediction accuracy for the DI trait in Labrador Retrievers, spanning from 0.7 to 0.9, slightly surpassing the prediction accuracies unearthed within our Sapsaree Population. Nonetheless, it’s imperative to acknowledge that prediction accuracy could potentially improve with enhanced linkage disequilibrium (LD) coverage achieved through the incorporation of more dense markers [47].
The observed increase in accuracy can be attributed to the simultaneous utilization of pedigree, phenotypic, and genomic data within the single-step method. This comprehensive approach provides a more comprehensive dataset for estimating breeding values compared to conventional pedigree-based models, which predominantly focus on capturing Mendelian sampling variations [48]. Furthermore, the ssGBLUP methodology exhibits a higher average accuracy gain across traits compared to the GBLUP approach, which solely relies on genotyped animal data. This disparity in accuracy likely stems from the incorporation of additional phenotypic data from non-genotyped animals, in conjunction with pedigree information, within the ssGBLUP framework [49].
On average, the GBLUP strategy exhibits marginally lower accuracy compared to the PBLUP approach, a phenomenon influenced by various factors. One contributing factor could be the limited number of genotyped reference animals in the GBLUP approach. Additionally, the PBLUP method utilizes all available phenotypic records and pedigree information from multiple generations to predict breeding values, contrasting with the GBLUP model’s reliance solely on data from genotyped animals in the current generation.
Our empirical findings underscore the superior performance of the GBLUP method relative to the PBLUP-G model, despite both utilizing identical phenotypic data. This aligns with previous research by Lee et al. [50], affirming the tendency for methods predicated solely on genomic information from genotyped animals to exhibit superior performance over PBLUP and PBLUP-G. Consequently, the GBLUP methodology emerges as particularly advantageous in scenarios where pedigree information is scarce. Furthermore, our results support prior studies indicating the superior accuracy of the ssGBLUP method over the PBLUP model. This comprehensive approach, which integrates both pedigree and genomic information, has consistently been shown to outperform pedigree-based BLUP or GBLUP methods in various contexts. In general, our findings suggest that the ssGBLUP method holds promise for improving prediction accuracy in practical breeding programs for Sapsaree traits.
The analysis of bias estimates reveals critical insights into the effectiveness of different prediction models in estimating breeding values. Our findings indicate that PBLUP and PBLUP-G show more pronounced bias across all traits compared to GBLUP and ssGBLUP. This trend raises important questions about the reliability of pedigree-based methods, particularly in contexts where accurate genetic predictions are essential for making informed breeding decisions. The observed bias in PBLUP and PBLUP-G suggests that these models may be limited in their ability to capture the true genetic relationships among individuals, particularly in complex traits where genomic information is paramount [51]. In contrast, GBLUP and ssGBLUP leverage genomic data more effectively, leading to lower bias estimates and potentially more accurate predictions. This difference highlights the advantages of integrating genomic information in breeding programs, as it can improve the accuracy of genetic evaluations and enhance overall selection strategies.
Moreover, while the average bias estimates provide a useful overview, they may mask important nuances present in specific traits. For instance, traits with high heritability might show different bias patterns than those with lower heritability [52]. This complexity emphasizes the need for a nuanced approach in evaluating prediction models, as biases may vary significantly depending on the trait being assessed. Therefore, it is critical to analyze individual traits in greater detail to gain a comprehensive understanding of each model’s strengths and weaknesses. The implications of these bias estimates extend beyond statistical considerations; they have practical significance in breeding decisions [23]. Models that consistently yield biased estimates can lead to suboptimal selection choices, ultimately hindering genetic progress. As such, breeders should be cautious in interpreting breeding values derived from models with higher bias, particularly when making decisions that impact long-term breeding strategies.
Indigenous breeds of domestic animals serve as invaluable resources for unraveling the intricate molecular mechanisms that underlie their diverse phenotypic traits, which have evolved through adaptation to either natural or human-induced changes. Numerous GWAS have successfully identified genes or genomic loci significantly associated with various morphological traits in dogs, shedding light on the pivotal role of genetic variations in shaping phenotypic diversity [53–55]. In our investigation, two positional candidate genes associated with body height, namely CCKAR and DCAF16, were identified on CFA3 (Table 9). The CCKAR (Cholecystokinin A Receptor) gene has been previously detected in Qinchuan beef cattle and is associated with growth and development, particularly in the small intestine, adipose tissue, muscle, and liver, while also regulating feeding behavior [56, 57]. Furthermore, polymorphisms in the promoter region of CCKAR have been correlated with fat deposition in humans [58]. Additionally, the CCKAR gene plays a crucial role in growth and body weight regulation during the domestication of chickens [57]. Similarly, the DCAF16 (DDB1 And CUL4 Associated Factor 16) gene has been associated with growth and carcass traits, including spleen weight [59], birth weight [60], bone weight [61], carcass weight [62], and average daily gain [63] in cattle. It also exhibits potential functions related to body weight in sheep [64] and body measurement traits in cattle [59].
The most significant candidate genes identified for body length are PDZRN3, located on CFA20, and CNTN1, located on CFA27 (Table 9). The PDZRN3 (PDZ domain-containing RING finger 3) gene has been associated with various functions including horse performance [65], heart maturation [66], and perturbation of skeletal muscle growth and maturation in transgenic mice, affecting the neuromuscular junction [67]. Conversely, deficiency in CNTN1 (Contactin 1) leads to animals developing normally until approximately P10, after which they fail to thrive, exhibiting symptoms such as ataxia and anorexia, and typically succumb between P16 and P18 [68]. Although the failure to thrive phenotype in CNTN1-deficient animals suggests intestinal malabsorption, various dietary interventions failed to improve health or extend lifespan [69].
For chest width, three genes were identified within the chromosomal regions of CFA2, CFA6, and CFA12. These genes are TRIM63, KDELR2, and SUPT3H (Table 9). The gene TRIM63 (Tripartite Motif Containing 63) plays an important role during skeletal muscle atrophy [70]. Recently discovered variants of the KDELR2 (KDEL Endoplasmic Reticulum Protein Retention Receptor 2) gene have been associated with multiple fractures beginning in childhood, long bone bowing, chest deformity, and short stature, leading to a diagnosis of progressive deforming osteogenesis imperfecta [71]. Furthermore, in a previous study, the SUPT3H (SPT3 Homolog, SAGA and STAGA Complex Component) gene was linked to bone and cartilage-related phenotypes, including height [72], bone mineral density [73], ossification of the posterior longitudinal ligament of the spine [74], and hip osteoarthritis [75].
The Sapsaree breed, indigenous to the Korean peninsula, boasts a rich history spanning several centuries, despite its genetic divergence from Western dog breeds [5, 76, 77]. Interestingly, Sapsaree carries an allele associated with the hair phenotype observed in European dog breeds, underscoring intriguing genetic connections [78]. The genes RSPO2, EIF3E, PKHD1L1, TRPS1, and EXT1, located on CFA13 8.40 Mb to 17.34 Mb, have been identified for hair length in the Sapsaree population (Table 9). The RSPO2 (R-Spondin 2) gene has been previously identified for the hair length of the entire body depending on the genetic background in the Korean Sapsaree dog population [79]. Currently, 197 long-haired dog breeds have been officially recognized by the American Kennel Club (AKC, www.ack.org/dog-breeds; accessed on 26 September 2021), and their hair length variations have been explained by the FGF5 and RSPO2 variations. The presence of the identical RSPO2-repeat allele among the RSPO2-associated long-haired dog breeds, including Sapsarees, suggests a common origin of the allele or an ancestral effect in long-haired dog breeds. This suggests that the genetic origin of hair length phenotypes could be different from the genetic background estimated from whole genomes. In a recent study, the EIF3E (Eukaryotic Translation Initiation Factor 3 Subunit E) gene is reported to be associated with the form of dermal papilla cells in the hair follicle regeneration of cashmere goats [80]. PKHD1L1 (PKHD1 Like 1) gene is predicted to encode a very large type-I transmembrane protein conserved function with a mammalian hair cell [81]. The TRPS1 (Transcriptional Repressor GATA Binding 1) gene has been suggested as a strong candidate gene for hair development [82], developing hair follicles, and is associated with hypertrichosis [83]. In a previous study, Zhang [82] reported that the TRPS1 expression was essential for normal hair follicle growth using TRPS1 mice. In addition, a recent study of comprehensive transcriptome profiling of balding and non-balding scalps in TRPS1 patients demonstrated that TRPS1 indeed plays a vital role in human hair loss [84]. Specifically, TRPS1 controls gene expression and plays a fundamental role in the interaction between epithelial and dermal papilla cells during hair follicle morphogenesis. It has been suggested that the EXT1 (Exostosin Glycosyltransferase 1) gene has been involved in various biological functional roles in humans such as large pear-shaped noses, thick and broad eyebrows, prominent ears, short stature, short digits, cone-shaped epiphyses, dystrophic nails, and fine, sparse hair [85]. In addition, in a mice model experiment, it is also evident that this gene also accelerated hair growth, and plays an important role in both hair follicle morphogenesis and homeostasis [86]. Thus, these genes collectively contribute to the intricate regulation of hair length and follicle development in the Sapsaree population, shedding light on the underlying genetic mechanisms governing these phenotypic traits.
Regarding the DI trait, a significant SNP was detected within the DDHD1, BMP4, SEMA3C, and FOXP1 genes located on CFA8, CFA18, and CFA20 (Table 9). The DDHD1 (DDHD Domain Containing 1) gene is implicated in a rare neurological disorder termed hereditary spastic paraplegia (HSP) subtype 28 (SPG28) in humans. This condition manifests with symptoms such as spastic gait, hyperreflexia, mild peripheral neuropathy, cerebellar eye movements, and urinary incontinence [87, 88]. Moreover, mutations in DDHD1 have been associated with juvenile ALS and neurodegeneration with brain iron accumulation [89]. Previous research has indicated that BMP4 (Bone morphogenetic protein 4) gene expression may contribute to various organ abnormalities and skeletal defects, including abnormal digit patterning and defects in mandibular development [90, 91]. Similarly, SEMA3C (Semaphorin 3C) gene variants have been linked to the regulation of the enteric nervous system in rodents and rare cases of Hirschsprung’s disease in humans [92]. Furthermore, the FOXP1 (Forkhead Box P1) gene has been implicated in several conditions, including intellectual disability, autism spectrum disorder, and congenital anomalies such as dysmorphic features and cardiac abnormalities [93].
Conclusion
In conclusion, this study sheds light on the accuracy of breeding values for various traits in Korean Sapsaree dogs and identifies genomic regions and candidate genes influencing these traits. Through comprehensive analyses of phenotypic and genotypic data, significant heritability estimates were obtained, underscoring the genetic basis of these traits. Notably, the ssGBLUP method exhibited the highest accuracy, suggesting its efficacy in assessing breeding values in canine populations. Furthermore, the identification of significant SNPs and candidate genes provides valuable insights into the genetic architecture of key traits in Sapsaree dogs. Notable genes associated with body height, body length, chest width, hair length, and distraction index were identified, paving the way for targeted breeding strategies and trait improvement initiatives. While these findings offer promising avenues for genetic improvement within the Sapsaree population, it is crucial to acknowledge certain limitations. Further validation of these results using high-density SNP chips and larger sample sizes is necessary to achieve a comprehensive understanding of trait expression and inheritance patterns. Overall, this study contributes to our understanding of canine genetics and provides a foundation for future research aimed at enhancing the health, welfare, and breed standards of Sapsaree dogs through informed breeding practices.
Supporting information
S1 Table. GO terms from DAVID software significantly enriched using candidate genes associated with the studied Sapsaree traits.
https://doi.org/10.1371/journal.pone.0312583.s001
(DOCX)
References
- 1. Galibert F.; Quignon P.; Hitte C.; Andre C. Toward understanding dog evolutionary and domestication history. C R Biol 2011, 334, 190–196, pmid:21377613
- 2. Wang G.D.; Zhai W.; Yang H.C.; Wang L.; Zhong L.; Liu Y.H.; et al. Out of southern East Asia: the natural history of domestic dogs across the world. Cell Res 2016, 26, 21–33, pmid:26667385
- 3.
Salmon, P.W.; Salmon, I.M. Society with animals: who owns who? Psychological research into the human-pet bond in Australia; University of Pennsylvania Press: Philadelphia, PA, 1983.
- 4. Ha J.H.; Alam M.; Lee D.H.; Kim J.J. Whole Genome Association Study to Detect Single Nucleotide Polymorphisms for Behavior in Sapsaree Dog (Canis familiaris). Asian-Australas J Anim Sci 2015, 28, 936–942, pmid:26104397
- 5. Gajaweera C.; Kang J.M.; Lee D.H.; Lee S.H.; Kim Y.K.; Wijayananda H.I.; et al. Genetic diversity and population structure of the Sapsaree, a native Korean dog breed. BMC Genet 2019, 20, 66, pmid:31382890
- 6. Cho G.J. Microsatellite polymorphism and genetic relationship in dog breeds in Korea. Asian Austral J Anim 2005, 18, 1071–1074.
- 7. Yoo D.; Kim K.; Kim H.; Cho S.; Kim J.N.; Lim D et al. The Genetic Origin of Short Tail in Endangered Korean Dog, DongGyeongi. Sci Rep 2017, 7, 10048, pmid:28855671
- 8. Raza S.H.A.; Khan S.; Amjadi M.; Abdelnour S.A.; Ohran H.; Alanazi K.M.; et al. Genome-wide association studies reveal novel loci associated with carcass and body measures in beef cattle. Arch Biochem Biophys 2020, 694, 108543, pmid:32798459
- 9. Haque M.A.; Alam M.Z.; Iqbal A.; Lee Y.M.; Dang C.G.; Kim J.J. Evaluation of accuracies of genomic predictions for body conformation traits in Korean Holstein. Animal bioscience 2024, 37, 555–566, pmid:38271974
- 10.
Schnelle, G.B. Some new diseases in dog. American Kennel Gazette 1935, pp. 25–26.
- 11. Wilson B.J.; Nicholas F.W.; James J.W.; Wade C.M.; Tammen I.; Raadsma H.W.; et al. Heritability and phenotypic variation of canine hip dysplasia radiographic traits in a cohort of Australian German shepherd dogs. PLoS One 2012, 7, e39620, pmid:22761846
- 12. Everts R.E.; Hazewinkel H.A.; Rothuizen J.; van Oost B.A. Bone disorders in the dog: a review of modern genetic strategies to find the underlying causes. Vet Q 2000, 22, 63–70, pmid:10789512
- 13. Ginja M.M.; Silvestre A.M.; Gonzalo-Orden J.M.; Ferreira A.J. Diagnosis, genetic control and preventive management of canine hip dysplasia: a review. Vet J 2010, 184, 269–276, pmid:19428274
- 14. Santana A.; Alves-Pimenta S.; Franco-Goncalo P.; Goncalves L.; Martins J.; Colaco B.; et al. Early hip laxity screening and later canine hip dysplasia development. Vet World 2022, 15, 679–684, pmid:35497953
- 15. Mei C.; Junjvlieke Z.; Raza S.H.A.; Wang H.; Cheng G.; Zhao C.; et al. Copy number variation detection in Chinese indigenous cattle by whole genome sequencing. Genomics 2020, 112, 831–836, pmid:31145994
- 16. Alam M.; Han K.I.; Lee D.H.; Ha J.H.; Kim J.J. Estimation of Effective Population Size in the Sapsaree: A Korean Native Dog (Canis familiaris). Asian-Australas J Anim Sci 2012, 25, 1063–1072, pmid:25049664
- 17.
Boichard, D. Pedig: a fortran package for pedigree analysis suited to large populations. In Proceedings of the 7th World Congress on Genetics Applied to Livestock Production, Montpellier, France, 19–23 août, 2002.
- 18.
Sargolzaei, M.; Iwaisaki, H.; Colleau, J. CFC: A tool for monitoring genetic diversity. In Proceedings of the In: Proceedings of the 8th world congress on genetics applied to livestock production, Belo Horizonte, MG, Brasil, August 13–18, 2006; pp. 27–28.
- 19. Purcell S.; Neale B.; Todd-Brown K.; Thomas L.; Ferreira M.A.; Bender D.; et al. PLINK: a tool set for whole-genome association and population-based linkage analyses. Am J Hum Genet 2007, 81, 559–575, pmid:17701901
- 20. Browning B.L.; Zhou Y.; Browning S.R. A One-Penny Imputed Genome from Next-Generation Reference Panels. Am J Hum Genet 2018, 103, 338–348, pmid:30100085
- 21.
Misztal, I.; Tsuruta, S.; Lourenco, D.; Masuda, Y.; Aguilar, I.; Legarra, A.; et al. Manual for BLUPF90 family of programs University of Georgia: Athens, USA, 2022.
- 22.
Gilmour, A.R.; Gogel, B.J.; Cullis, B.R.; Welham, S.J.; Thompson, R. ASReml User Guide Release 4.2 Functional Specification, VSN International Ltd: Hemel Hempstead, HP2 4TP, UK, 2021.
- 23. Haque M.A.; Lee Y.-M.; Ha J.-J.; Jin S.; Park B.; Kim N.-Y.; et al. Genomic Predictions in Korean Hanwoo Cows: A Comparative Analysis of Genomic BLUP and Bayesian Methods for Reproductive Traits. Animals-Basel 2024, 14, 27, pmid:38200758
- 24. Haque M.A.; Iqbal A.; Bae H.; Lee S.E.; Park S.; Lee Y.M.; et al. Assessment of genomic breeding values and their accuracies for carcass traits in Jeju Black cattle using whole-genome SNP chip panels. J Anim Breed Genet 2023, 140, 519–531, pmid:37102238
- 25.
Lynch, M.; Walsh, B. Genetics and analysis of quantitative traits; Sinauer Associates Inc.: Massachuetts, USA, 1998; p. 1008.
- 26. VanRaden P.M. Efficient methods to compute genomic predictions. J Dairy Sci 2008, 91, 4414–4423, pmid:18946147
- 27. Aguilar I.; Misztal I.; Johnson D.L.; Legarra A.; Tsuruta S.; Lawlor T.J. Hot topic: a unified approach to utilize phenotypic, full pedigree, and genomic information for genetic evaluation of Holstein final score. J Dairy Sci 2010, 93, 743–752, pmid:20105546
- 28.
Aguilar, I.; Misztal, I.; Tsuruta, S.; Legarra, A.; Wang, H. PREGSF90 –POSTGSF90: Computational tools for the implementation of single-step genomic selection and genome-wide association with ungenotyped individuals in BLUPF90 programs. Proceedings of the World Congress on Genetics Applied to Livestock Production 2014, Methods and Tools: Statistical and genomic tools for mapping QTL and genes (Posters), 680.
- 29. Zhou X.; Stephens M. Genome-wide efficient mixed-model analysis for association studies. Nat Genet 2012, 44, 821–824, pmid:22706312
- 30. Haque M.A.; Alam M.Z.; Iqbal A.; Lee Y.-M.; Dang C.-G.; Kim J.-J. Genome-Wide Association Studies for Body Conformation Traits in Korean Holstein Population. 2023, 13, 2964.
- 31.
R Core Team R: A language and environment for statistical computing, R Foundation for Statistical Computing: Vienna, Austria, 2023.
- 32. Huang D.W.; Sherman B.T.; Tan Q.; Collins J.R.; Alvord W.G.; Roayaei J.; et al. The DAVID Gene Functional Classification Tool: a novel biological module-centric algorithm to functionally analyze large gene lists. Genome Biol 2007, 8, R183, pmid:17784955
- 33. Xie C.; Mao X.; Huang J.; Ding Y.; Wu J.; Dong S.; et al. KOBAS 2.0: a web server for annotation and identification of enriched pathways and diseases. Nucleic Acids Res 2011, 39, W316–322, pmid:21715386
- 34. Guo Y.; Huang Y.; Hou L.; Ma J.; Chen C.; Ai H.; et al. Genome-wide detection of genetic markers associated with growth and fatness in four pig populations using four approaches. Genet Sel Evol 2017, 49, 21, pmid:28196480
- 35. Haque M.A.; Alam M.Z.; Iqbal A.; Lee Y.M.; Dang C.G.; Kim J.J. Genome-Wide Association Studies for Body Conformation Traits in Korean Holstein Population. Animals (Basel) 2023, 13, 2964, pmid:37760364
- 36. Verryn S.D.; Geerthsen J.M.P. Heritabilities of a population of German Shepherd Dogs with a complex interrelationship structure. Theoretical and Applied Genetics 1987, 75, 144–146,
- 37.
Famula, T.R. Genetics of quantitative traits and improvement of dog breeds; Sampson, A.R.a.J., Ed.; CAB International: Wallingford, Oxon, UK, 2001.
- 38. Ginja M.M.; Silvestre A.M.; Ferreira A.J.; Gonzalo-Orden J.M.; Orden M.A.; Melo-Pinto P.; et al. Passive hip laxity in Estrela Mountain Dog—distraction index, heritability and breeding values. Acta Vet Hung 2008, 56, 303–312, pmid:18828482
- 39. Todhunter R.J.; Bliss S.P.; Casella G.; Wu R.; Lust G.; Burton-Wurster N.I.; et al. Genetic structure of susceptibility traits for hip dysplasia and microsatellite informativeness of an outcrossed canine pedigree. J Hered 2003, 94, 39–48, pmid:12692161
- 40. Zhang Z.; Zhu L.; Sandler J.; Friedenberg S.S.; Egelhoff J.; Williams A.J.; et al. Estimation of heritabilities, genetic correlations, and breeding values of four traits that collectively define hip dysplasia in dogs. Am J Vet Res 2009, 70, 483–492, pmid:19335104
- 41. Leighton E.A.; Holle D.; Biery D.N.; Gregor T.P.; McDonald-Lynch M.B.; Wallace M.L.; et al. Genetic improvement of hip-extended scores in 3 breeds of guide dogs using estimated breeding values: Notable progress but more improvement is needed. PLoS One 2019, 14, e0212544, pmid:30794614
- 42.
Tikekar, A.; Soo, M.; Erceg, V.H.; Villalobos, N.L.; Worth, A.J. Estimate of heritability of the PennHIP distraction index in working German shepherd dogs in New Zealand. In Proceedings of the 7th International Veterinary Congress, Paris, France, 2017.
- 43. Tikekar A.; Soo M.; Lopez-Villalobos N.; Worth A.J. Provisional heritability estimates of four distraction index traits in a breeding population of German Shepherd dogs. N Z Vet J 2018, 66, 319–324, pmid:30122125
- 44. Wierzbicki H. Breeding value evaluation in Polish fur animals: Estimates of direct heritability and portion of litter variation of fur coat and reproduction traits. Czech Journal of Animal Science 2004, 49, 474–482,
- 45. Ruefenacht S.; Gebhardt-Henrich S.; Miyake T.; Gaillard C. A behaviour test on German Shepherd dogs: heritability of seven different traits. Applied Animal Behaviour Science 2002, 79, 113–132,
- 46. Guo G.; Zhou Z.; Wang Y.; Zhao K.; Zhu L.; Lust G.; et al. Canine hip dysplasia is predictable by genotyping. Osteoarthritis Cartilage 2011, 19, 420–429, pmid:21215318
- 47. Habier D.; Fernando R.L.; Dekkers J.C. Genomic selection using low-density marker panels. Genetics 2009, 182, 343–353, pmid:19299339
- 48. Christensen O.F.; Madsen P.; Nielsen B.; Ostersen T.; Su G. Single-step methods for genomic evaluation in pigs. Animal 2012, 6, 1565–1571, pmid:22717310
- 49. Naserkheil M.; Lee D.H.; Mehrban H. Improving the accuracy of genomic evaluation for linear body measurement traits using single-step genomic best linear unbiased prediction in Hanwoo beef cattle. BMC Genet 2020, 21, 144, pmid:33267771
- 50. Lee J.; Cheng H.; Garrick D.; Golden B.; Dekkers J.; Park K.; et al. Comparison of alternative approaches to single-trait genomic prediction using genotyped and non-genotyped Hanwoo beef cattle. Genet Sel Evol 2017, 49, 2, pmid:28093065
- 51. Vallejo R.L.; Cheng H.; Fragomeni B.O.; Gao G.T.; Silva R.M.O.; Martin K.E.; et al. The accuracy of genomic predictions for bacterial cold water disease resistance remains higher than the pedigree-based model one generation after model training in a commercial rainbow trout breeding population. Aquaculture 2021, 545, 737164,
- 52. Polderman T.J.; Benyamin B.; de Leeuw C.A.; Sullivan P.F.; van Bochoven A.; Visscher P.M.; et al. Meta-analysis of the heritability of human traits based on fifty years of twin studies. Nat Genet 2015, 47, 702–709, pmid:25985137
- 53. Plassais J.; Kim J.; Davis B.W.; Karyadi D.M.; Hogan A.N.; Harris A.C.; et al. Whole genome sequencing of canids reveals genomic regions under selection and variants influencing morphology. Nat Commun 2019, 10, 1489, pmid:30940804
- 54. Hayward J.J.; Castelhano M.G.; Oliveira K.C.; Corey E.; Balkman C.; Baxter T.L.; et al. Complex disease and phenotype mapping in the domestic dog. Nat Commun 2016, 7, 10460, pmid:26795439
- 55. Schoenebeck J.J.; Ostrander E.A. Insights into morphology and disease from the dog genome project. Annu Rev Cell Dev Biol 2014, 30, 535–560, pmid:25062362
- 56. Nurgulsim K.; Raza S.H.A.; Khan R.; Shah M.A.; Jahejo A.R.; Batool U.; et al. Identification of genetic variants the CCKAR gene and based on bodymeasurement and carcass quality characteristics in Qinchuan beef cattle (Bos taurus). Electronic Journal of Biotechnology 2021, 51, 1–7,
- 57. Dunn I.C.; Meddle S.L.; Wilson P.W.; Wardle C.A.; Law A.S.; Bishop V.R.; et al. Decreased expression of the satiety signal receptor CCKAR is responsible for increased growth and body weight during the domestication of chickens. American journal of physiology. Endocrinology and metabolism 2013, 304, E909–921, pmid:23443924
- 58. Funakoshi A.; Miyasaka K.; Matsumoto H.; Yamamori S.; Takiguchi S.; Kataoka K.; et al. Gene structure of human cholecystokinin (CCK) type-A receptor: body fat content is related to CCK type-A receptor gene promoter polymorphism. FEBS Lett 2000, 466, 264–266, pmid:10682840
- 59. An B.; Xia J.; Chang T.; Wang X.; Xu L.; Zhang L.; et al. Genome-wide association study reveals candidate genes associated with body measurement traits in Chinese Wagyu beef cattle. Anim Genet 2019, 50, 386–390, pmid:31179577
- 60. Smith J.L.; Wilson M.L.; Nilson S.M.; Rowan T.N.; Oldeschulte D.L.; Schnabel R.D.; et al. Genome-wide association and genotype by environment interactions for growth traits in U.S. Gelbvieh cattle. BMC genomics 2019, 20, 926, pmid:31801456
- 61. An B.; Gao X.; Chang T.; Xia J.; Wang X.; Miao J.; et al. Genome-wide association studies using binned genotypes. Heredity (Edinb) 2020, 124, 288–298, pmid:31641238
- 62. Bhuiyan M.S.A.; Lim D.; Park M.; Lee S.; Kim Y.; Gondro C.; et al. Functional Partitioning of Genomic Variance and Genome-Wide Association Study for Carcass Traits in Korean Hanwoo Cattle Using Imputed Sequence Level SNP Data. Front Genet 2018, 9, 217, pmid:29988410
- 63. Zhang W.; Li J.; Guo Y.; Zhang L.; Xu L.; Gao X.; et al. Multi-strategy genome-wide association studies identify the DCAF16-NCAPG region as a susceptibility locus for average daily gain in cattle. Sci Rep 2016, 6, 38073, pmid:27892541
- 64. La Y.; Zhang X.; Li F.; Zhang D.; Li C.; Mo F.; et al. Molecular Characterization and Expression of SPP1, LAP3 and LCORL and Their Association with Growth Traits in Sheep. Genes (Basel) 2019, 10, pmid:31416156
- 65.
Littiere, T.O.; Castro, G.H.F.; Rodriguez, M.d.P.R.; Bonafé, C.M.; Magalhães, A.F.B.; Faleiros, R.R.; et al. Identification and Functional Annotation of Genes Related to Horses’ Performance: From GWAS to Post-GWAS. 2020, 10, 1173.
- 66. Pernot M.; Jaspard-Vinassa B.; Abelanet A.; Rubin S.; Forfar I.; Jeanningros S.; et al. Decrease of Pdzrn3 is required for heart maturation and protects against heart failure. Sci Rep 2022, 12, 8, pmid:34996942
- 67. Lu Z.; Je H.S.; Young P.; Gross J.; Lu B.; Feng G. Regulation of synaptic growth and maturation by a synapse-associated E3 ubiquitin ligase at the neuromuscular junction. The Journal of cell biology 2007, 177, 1077–1089, pmid:17576800
- 68. Fetissov S.O.; Bergstrom U.; Johansen J.E.; Hokfelt T.; Schalling M.; Ranscht B. Alterations of arcuate nucleus neuropeptidergic development in contactin-deficient mice: comparison with anorexia and food-deprived mice. The European journal of neuroscience 2005, 22, 3217–3228, pmid:16367788
- 69. Veny M.; Grases D.; Kucharova K.; Lin W.W.; Nguyen J.; Huang S.; et al. Contactin-1 Is Required for Peripheral Innervation and Immune Homeostasis Within the Intestinal Mucosa. Frontiers in immunology 2020, 11, 1268, pmid:32676079
- 70. Peris-Moreno D.; Taillandier D.; Polge C. MuRF1/TRIM63, Master Regulator of Muscle Mass. Int J Mol Sci 2020, 21, pmid:32933049
- 71. van Dijk F.S.; Semler O.; Etich J.; Kohler A.; Jimenez-Estrada J.A.; Bravenboer N.; et al. Interaction between KDELR2 and HSP47 as a Key Determinant in Osteogenesis Imperfecta Caused by Bi-allelic Variants in KDELR2. Am J Hum Genet 2020, 107, 989–999, pmid:33053334
- 72. Lango Allen H.; Estrada K.; Lettre G.; Berndt S.I.; Weedon M.N.; Rivadeneira F.; et al. Hundreds of variants clustered in genomic loci and biological pathways affect human height. Nature 2010, 467, 832–838, pmid:20881960
- 73. Estrada K.; Styrkarsdottir U.; Evangelou E.; Hsu Y.H.; Duncan E.L.; Ntzani E.E.; et al. Genome-wide meta-analysis identifies 56 bone mineral density loci and reveals 14 loci associated with risk of fracture. Nat Genet 2012, 44, 491–501, pmid:22504420
- 74. Nakajima M.; Takahashi A.; Tsuji T.; Karasugi T.; Baba H.; Uchida K.; et al. A genome-wide association study identifies susceptibility loci for ossification of the posterior longitudinal ligament of the spine. Nat Genet 2014, 46, 1012–1016, pmid:25064007
- 75. Castano-Betancourt M.C.; Evans D.S.; Ramos Y.F.; Boer C.G.; Metrustry S.; Liu Y.; et al. Novel Genetic Variants for Cartilage Thickness and Hip Osteoarthritis. PLoS Genet 2016, 12, e1006260, pmid:27701424
- 76. Jeong H.; Choi B.H.; Eo J.; Kwon Y.J.; Lee H.E.; Choi Y.; et al. Statistical analysis and genetic diversity of three dog breeds using simple sequence repeats. Genes & Genomics 2014, 36, 883–889,
- 77. Kang M.; Ahn B.; Youk S.; Cho H.S.; Choi M.; Hong K.; et al. High Allelic Diversity of Dog Leukocyte Antigen Class II in East Asian Dogs: Identification of New Alleles and Haplotypes. Journal of Mammalian Evolution 2021, 28, 773–784,
- 78. Parker H.G.; Chase K.; Cadieu E.; Lark K.G.; Ostrander E.A. An insertion in the RSPO2 gene correlates with improper coat in the Portuguese water dog. J Hered 2010, 101, 612–617, pmid:20562213
- 79. Kang M.; Ahn B.; Youk S.; Lee Y.M.; Kim J.J.; Ha J.H.; et al. Tracing the Origin of the RSPO2 Long-Hair Allele and Epistatic Interaction between FGF5 and RSPO2 in Sapsaree Dog. Genes (Basel) 2022, 13, 102, pmid:35052442
- 80. Yang F.; Li R.; Zhao C.; Che T.; Guo J.; Xie Y.; et al. Single-cell sequencing reveals the new existence form of dermal papilla cells in the hair follicle regeneration of cashmere goats. Genomics 2022, 114, 110316, pmid:35202721
- 81. Makrogkikas S.; Cheng R.K.; Lu H.; Roy S. A conserved function of Pkhd1l1, a mammalian hair cell stereociliary coat protein, in regulating hearing in zebrafish. J Neurogenet 2023, 37, 85–92, pmid:36960824
- 82. Zhang Y.; Nakamura T.; Furukawa F.; Muragaki Y. Trps1-deficient transplanted skin gave rise to a substantial amount of hair: Trps1 is unnecessary for hair development. Dermatology reports 2019, 11, 7853, pmid:30815242
- 83. Fantauzzo K.A.; Kurban M.; Levy B.; Christiano A.M. Trps1 and its target gene Sox9 regulate epithelial proliferation in the developing hair follicle and are associated with hypertrichosis. PLoS Genet 2012, 8, e1003002, pmid:23133399
- 84. Kim Y.J.; Yoon B.; Han K.; Park B.C. Comprehensive Transcriptome Profiling of Balding and Non-Balding Scalps in Trichorhinophalangeal Syndrome Type I Patient. Annals of dermatology 2017, 29, 597–601, pmid:28966516
- 85. Ahmed A.; Almohanna H.; Griggs J.; Tosti A. Genetic Hair Disorders: A Review. Dermatology and therapy 2019, 9, 421–448, pmid:31332722
- 86. Coulson-Thomas V.J.; Gesteira T.F.; Esko J.; Kao W. Heparan sulfate regulates hair follicle and sebaceous gland morphogenesis and homeostasis. J Biol Chem 2014, 289, 25211–25226, pmid:25053416
- 87. Mignarri A.; Rubegni A.; Tessa A.; Stefanucci S.; Malandrini A.; Cardaioli E.; et al. Mitochondrial dysfunction in hereditary spastic paraparesis with mutations in DDHD1/SPG28. Journal of the neurological sciences 2016, 362, 287–291, pmid:26944165
- 88. Liguori R.; Giannoccaro M.P.; Arnoldi A.; Citterio A.; Tonon C.; Lodi R.; et al. Impairment of brain and muscle energy metabolism detected by magnetic resonance spectroscopy in hereditary spastic paraparesis type 28 patients with DDHD1 mutations. Journal of neurology 2014, 261, 1789–1793, pmid:24989667
- 89. Dard R.; Meyniel C.; Touitou V.; Stevanin G.; Lamari F.; Durr A.; et al. Mutations in DDHD1, encoding a phospholipase A1, is a novel cause of retinopathy and neurodegeneration with brain iron accumulation. European journal of medical genetics 2017, 60, 639–642, pmid:28818478
- 90. Wang R.N.; Green J.; Wang Z.; Deng Y.; Qiao M.; Peabody M.; et al. Bone Morphogenetic Protein (BMP) signaling in development and human diseases. Genes Dis 2014, 1, 87–105, pmid:25401122
- 91. Ren Y.; Feng J.; Liu Z.Z.; Wan H.; Li J.H.; Lin X. A new haplotype in BMP4 implicated in ossification of the posterior longitudinal ligament (OPLL) in a Chinese population. J Orthop Res 2012, 30, 748–756, pmid:22052794
- 92. van der Klaauw A.A.; Croizier S.; Mendes de Oliveira E.; Stadler L.K.J.; Park S.; Kong Y.; et al. Human Semaphorin 3 Variants Link Melanocortin Circuit Development and Energy Balance. Cell 2019, 176, 729–742 e718, pmid:30661757
- 93. Lozano R.; Gbekie C.; Siper P.M.; Srivastava S.; Saland J.M.; Sethuram S.; et al. FOXP1 syndrome: a review of the literature and practice parameters for medical assessment and monitoring. Journal of neurodevelopmental disorders 2021, 13, 18, pmid:33892622