Genome-Wide Analyses of Breed Differentiation and Functional Specialization in Domestic Dogs

Genome-Wide Analyses of Breed Differentiation and Functional Specialization in Domestic Dogs | Research Square window.SnipcartSettings = { analytics: { enabled: false } }; (function() { var accessVector = localStorage.getItem('access_vector') || ''; window.dataLayer = window.dataLayer || []; if (accessVector) { window.dataLayer.push({ user: { profile: { profileInfo: { snid: accessVector } } } }); } })(); (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-K279D39R'); Browse Preprints In Review Journals COVID-19 Preprints AJE Video Bytes Research Tools Research Promotion AJE Professional Editing AJE Rubriq About Preprint Platform In Review Editorial Policies Our Team Advisory Board Help Center Sign In Submit a Preprint Cite Share Download PDF Research Article Genome-Wide Analyses of Breed Differentiation and Functional Specialization in Domestic Dogs Rongxing Wei, jiaqian Le, Linying Ye, Linyuan Guo, Chang Su, Mingyue Zhao, and 5 more This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-9303752/v1 This work is licensed under a CC BY 4.0 License Status: Under Review Version 1 posted 12 You are reading this latest preprint version Abstract Domestic dog breeds exhibit extensive phenotypic and functional diversity shaped by long-term artificial selection. However, the genomic processes underlying lineage divergence and functional differentiation between working and companion dogs remain incompletely understood. Here, we analysed over 5,000 genome-wide SNP data to investigate population structure, evolutionary dynamics, and functional genomic differentiation across multiple dog breeds. Genome-wide dimensionality reduction revealed a star-like genetic structure radiating from gray wolves, indicating directional divergence of modern breeds along partially independent trajectories. We further identified nine major lineages broadly corresponding to functional breed groups. Genomic analyses showed that modern purebred dogs exhibit markedly elevated genomic inbreeding compared with village dogs and wolves, and population-level F ROH was strongly positively correlated with mean pairwise F ST among breeds, suggesting that intensive artificial breeding has simultaneously increased within-breed homozygosity and between-breed genetic divergence. Comparative analyses between working and companion dogs revealed significant enrichment of muscle-related pathways, including actin cytoskeleton, myofibril, and contractile muscle fiber, with multiple myosin family genes forming a distinct functional module. Finally, we developed a compact breed identification panel based on ancestry-informative SNPs, in which a 200-SNP panel accurately distinguished eight target breeds. These findings provide insights into the genomic basis of breed diversification and functional specialization in domestic dogs. Domestic dogs Population structure Genomic homozygosity Functional differentiation Ancestry-informative markers Figures Figure 1 Figure 2 Figure 3 Figure 4 Introduction Human-dog coexistence can be traced back approximately 14,000–32,000 years, representing one of the earliest and longest-standing relationships between humans and animals [ 1 – 4 ]. During the domestication process, dogs ( Canis lupus familiaris ) experienced extensive diversification under long-term human-mediated selection. Today, more than 450 breeds are officially recognized by major kennel organizations such as the American Kennel Club and the Fédération Cynologique Internationale [ 5 ]. This remarkable phenotypic diversity has made dogs an important model for comparative genomics and studies of phenotypic evolution. Genome-wide association studies have identified the genetic basis of numerous canine traits, including body size, coat color, limb length, skull morphology, and lifespan, highlighting the power of the dog model for dissecting genotype-phenotype relationships [ 6 – 9 ]. Additionally, the dog model has been utilized to identify genes with translational potential for human health and biology, including both rare and common human disorders, such as autoimmune disease, neuromuscular disorders, and cancer [ 10 – 15 ]. Recent ancient genomic studies further suggest that dogs and humans dispersed together across Eurasia, indicating a long history of co-dispersal between the two species [ 16 ]. In some cases, dogs possessing valuable working or physical traits were exchanged among human populations, facilitating the spread of functionally specialized lineages. However, most modern dog breeds emerged relatively recently, primarily during the mid-19th century, when deliberate reproductive isolation and intensive artificial selection led to the establishment of standardized breeds [ 17 – 20 ]. These processes, often accompanied by strong population bottlenecks and closed breeding systems, generated pronounced population structure and genetic differentiation among breeds. Functional specialization has therefore been a major driver of canine diversity [ 8 ]. Historically, different breeds were selectively bred for specific purposes, such as hunting, guarding, herding, and companionship [ 21 ]. Consequently, modern dog populations can be broadly categorized into working dogs and companion dogs. However, despite their clear functional divergence, the genetic differences between working and companion dogs remain insufficiently characterized. In particular, widely used police and working breeds, including German Shepherd Dog (GS), Belgian Malinois (BM), Kunming Dog (KM), Bloodhound (BH), Labrador Retriever (LR), Rottweiler (RT), and English Springer Spaniel (SS), as well as highly popular companion breeds such as the French Bulldog (FB), provide a valuable system for investigating the genomic basis of functional specialization in domestic dogs. In this study, we aimed to investigate the genetic basis of functional differentiation among dog breeds using genome-wide SNP data. Specifically, we characterized the population structure and genetic relationships among multiple dog breeds, identified genomic regions showing strong differentiation between working and companion dogs through genome-wide FST scans, and developed compact SNP panels capable of reliably distinguishing among target breeds. These analyses provide insights into the genomic signatures associated with functional divergence in dogs and offer practical tools for breed identification and breeding management. Material and methods Sample Collection Eight dog breeds were selected from the sample repository, including seven working breeds and one popular companion breed: German Shepherd Dog (GS), Belgian Malinois Shepherd Dog (BM), Kunming Dog (KM), Bloodhound (BH), Labrador Retriever (LR), Rottweiler (RT), English Springer Spaniel (SS), and French Bulldog (FB). According to the Fédération Cynologique Internationale (FCI) classification, GS and BM belong to Group 1 (Sheepdogs), RT to Group 2 (Molossoid breeds), BH to Group 6 (Scent hounds), LR and SS to Group 8 (Retrievers), and FB to Group 9 (Small Molossoid-type Dogs). KM is a Chinese native breed that is not recognized by the FCI. Samples were obtained from the Nanjing Police Dog Research Institute, Kunming Police Dog Base, and Nanchang Police Dog Base. A total of 312 individuals were included: 36 BM, 20 BH, 71 GS, 46 FB, 27 KM, 25 SS, 42 LR, and 45 RT. All sampled dogs were unrelated for at least two generations. DNA Extraction Genomic DNA was extracted from dried blood spots using a commercial DNA extraction kit following the manufacturer's protocol. Briefly, three 3×3 mm blood spot punches were incubated with lysis buffer and Proteinase K until fully solubilized. DNA was then bound to the column in the presence of ethanol, followed by sequential washes to remove impurities and residual salts. Columns were air-dried to eliminate residual ethanol, and DNA was finally eluted in 20–50 µL elution buffer for downstream analyses. SNP Genotyping and Quality Control Genomic DNA was genotyped using the Illumina Canine 230K SNP array following the manufacturer's protocol. DNA concentrations were measured and normalized to 50 ng/µL, with 200 ng used for the Infinium HD Assay. Genomic DNA was denatured and subjected to whole-genome amplification, yielding 2000-3000-fold increase in DNA amount. Amplified DNA was enzymatically fragmented in a controlled end-point reaction, precipitated with isopropanol, and resuspended in hybridization buffer. DNA samples were hybridized to the bead-based array overnight, allowing 50-base site-specific probes on each bead to anneal to the complementary DNA fragments. Non-specifically bound DNA was removed, followed by single-base extension and fluorescent labeling. Arrays were scanned using an Illumina scanner, and fluorescence intensities were converted to SNP genotypes using BeadStudio software. Genotyping quality was high, with a mean call rate of 96.57%. Ten individuals with call rates below 90% were excluded from further analyses. After applying additional filtering for missingness (--mind 0.01) and removing SNPs with a minor allele frequency below 0.05, a total of 290 individuals and 196,125 markers were retained for downstream analyses. Integration of Modern Genomic Data We further integrated publicly available genotype data from a global dataset, comprising 5,392 dogs [ 22 ]. Only SNPs genotyped on both arrays and passing quality control were included (69,018 SNPs). This dataset includes both breed dogs and village dogs sampled across multiple geographic regions, including Africa, America, Middle East, Australia, Central Asia, East Asia, Europe, and the Arctic. All modern breed dogs were classified according to the FCI grouping system. Population genetic analysis Dimensionality Reduction Principal component analysis (PCA) of unrelated modern dogs was performed using the smartpca program implemented in EIGENSOFT v3.0 [ 23 ]. The two-dimensional PHATE embedding was generated using the first 20 PCs as input with phate (pca_matrix, ndim = 2, knn = 80, gamma = 0, decay = 100) [ 24 ]. Slingshot v1.4.0 was used for identifying lineages using PHATE coordinates as input and clusters defined using kmeans (centers = 40) [ 25 ]. ADMIXTURE Analysis To estimate the ancestry of dogs, SNPs were pruned for linkage disequilibrium (LD) using PLINK v1.9 (--indep-pairwise 50 5 2) [ 26 ]. ADMIXTURE v1.3.0 was run on the pruned dataset for K = 2–20 with 10-fold cross-validation and 100 bootstrap replicates [ 27 ]. Cross-validation (CV) errors were calculated for each K to determine the optimal model, and the best-supported K was identified based on the lowest CV error. Ancestry proportions were visualized using AncestryPainterV2 [ 28 ]. Pairwise Fixation Index (FST) Estimation We then calculated pairwise fixation indexes (FST) among geographically distinct dogs using PLINK v1.9 with the --fst flag [ 26 , 29 ]. Runs of Homozygosity We used PLINK v1.9 to identify individual-level ROHs and calculated the population average level among geographically different dogs [ 26 ]. For each detected ROH segment, physical length (in kilobases) was converted to megabases (Mb). For each individual, the total length of ROH segments was calculated by summing all ROH segments across autosomes. Genomic inbreeding coefficients based on ROH (F_ROH) were calculated as: where \(\:\text{L=2203.76}\) Mb corresponds to the total length of the CanFam3.1 autosomal genome. Treemix Analysis Population trees and historical gene flow were inferred using Treemix v1.13 [ 30 ]. For each number of migration edges (m = 0–7), five replicates were run with the gray wolf as outgroup, 500-SNP blocks (-k 500), bootstrap support (-bootstrap), and standard errors of migration edges (-se). Global optimization (-global) was applied and sample size correction was disabled (-noss). Runs were executed in parallel in the background. The optimal number of migration edges was determined using OptM, and trees were visualized with custom R functions and exported as PDF files. Identification and functional characterization of highly differentiated loci Eight breeds were assigned to groups following the classification of the FCI. In addition, KM was incorporated into Group 1 because of its genetic affinity to GS and their similar working functions. Group 9 represents companion dogs, whereas all other groups correspond to working dogs. We calculated pairwise genetic differentiation (FST) between Group 9 and each of the working dog groups using a sliding window approach and obtained the mean FST for each window. Genomic regions within the top 1% of the empirical FST distribution were defined as highly differentiated loci. Genes located within these candidate regions were annotated based on the Ensembl gene annotation database (CanFam3.1, release 104). To explore the potential biological functions of the candidate genes, Gene Ontology enrichment analyses were conducted using the R package clusterProfiler. In addition, gene-concept network visualization was performed using the cnetplot function to illustrate the relationships between enriched functional categories and the associated genes. Finally, candidate gene sets identified from each group comparison were compared using Venn analysis to identify genes shared across groups. Identification of ancestry-informative SNP panels and breed classification using population structure and machine learning analyses Allele frequencies and pairwise linkage disequilibrium (LD) for each SNP were calculated in PLINK v1.9. To identify ancestry-informative markers capable of discriminating among the eight breeds, the software AIM-generator [ 31 ] was applied. Using this approach, three SNP panels containing 10, 100, and 200 markers were selected to evaluate their ability to distinguish among the eight dog breeds. In addition, PCA was conducted using the selected SNP sets with PLINK to visualize genetic clustering among individuals. Ancestry proportions were estimated using ADMIXTURE run assuming K = 8 ancestral populations, corresponding to the number of breeds analyzed. To further evaluate the effectiveness of the selected SNP panels for breed classification, machine learning models (Random Forest and XGBoost) were constructed based on the results of PCA and ADMIXTURE. The dataset was randomly divided into training (70%) and validation (30%) sets, while maintaining proportional representation of each breed. Hyperparameters for each model were optimized using grid search with cross-validation to maximize classification accuracy on the training dataset. The optimized models were then applied to the validation dataset, and classification performance was evaluated based on overall accuracy and confusion matrices. Results Genome-wide Dimensionality Reduction Identifies Dog Lineages Shaped by Divergent Working Roles All breed dogs in this study were categorized according to the Fédération Cynologique Internationale (FCI) classification system, with additional stratification based on their documented geographic origins to facilitate biologically meaningful comparisons. To investigate genetic relationships among individuals, we performed principal component analysis (PCA) using genome-wide SNP data. The two-dimensional projection of the first two principal components revealed robust genetic structuring across dog breeds (Fig. 1 A). Notably, the resulting pattern exhibited a star-like configuration radiating from gray wolves, which were positioned near the center of the plot. This central placement likely reflects their ancestral status relative to domestic dogs. Most European breeds were distributed along three well-defined clines extending outward from the wolf cluster, indicating directional divergence along partially independent genetic axes. Among the eight focal dog breeds, German Shepherd Dog (GS), Belgian Malinois Shepherd Dog (BM), Kunming Dog (KM), Labrador Retriever (LR), Rottweiler (RT), and French Bulldog (FB) were distinctly separated from gray wolves along the primary axis of variation, forming a tightly grouped cluster. The KM occupied an intermediate position along one of the European-associated clines, consistent with its breeding history, which involved admixture between German Shepherd Dogs imported from the Soviet Union and indigenous village dogs from Kunming in the 1950s [ 32 ]. In contrast, Bloodhound (BH) and English Springer Spaniel (SS) were positioned closer to the wolf cluster, reflecting reduced genetic differentiation relative to other breeds. To capture global relationships beyond discrete breed clusters, we applied potential of heat diffusion for affinity-based transition embedding (PHATE). PHATE preserves both local and global structure in high-dimensional data and models pairwise relationships among samples as diffusion probabilities. The resulting two-dimensional embedding displayed a star-shaped configuration with gray wolves near the center and multiple trajectories radiating outward. Trajectory extremities corresponded to the most genetically differentiated breeds, whereas basal portions occupied intermediate positions relative to the central wolf cluster. These patterns illustrate the relative genetic divergence and continuity of genomic variation among dog breeds. Mapping breeds according to the ten FCI groups revealed that breeds from different functional or geographic categories generally followed distinct trajectories within the PHATE manifold (Supplementary Fig. S1 ). For example, Group 10 breeds aligned along one trajectory, Group 5 along another, and freely breeding village dogs clustered near the center, consistent with lower differentiation. To formally characterize these patterns, we applied pseudotemporal reconstruction, which orders clusters of individuals along branching trajectories in high-dimensional space. Clusters were defined using k-means clustering, with gray wolves serving as a central reference. In this unsupervised framework, nine major lineages were identified, whereas breeds in Group 4 and Group 6 remained near the central region, reflecting their relatively low genetic differentiation compared to other modern breeds (Fig. 1 B). To further examine ancestral composition, we performed ADMIXTURE analysis at K = 15, which minimized cross-validation error. The results indicated that GS, RT, LR, and FB each possessed predominantly distinct ancestral components (Fig. 1 C). In contrast, KM, BH, BM, and ESS displayed more admixed patterns. Notably, the Kunming Dog genome comprised over 50 percent ancestry related to the German Shepherd Dog, whereas Bloodhounds, Belgian Shepherd Dogs, and English Springer Spaniels contained varying proportions of ancestry associated with the English Springer Spaniel ancestral component. Genomic Inbreeding and Genetic Drift Reinforce Breed Differentiation To investigate the demographic and evolutionary processes underlying the lineage structure identified above, we examined patterns of runs of homozygosity (ROH) within breeds, genome-wide genetic differentiation among breeds, and historical gene flow using TreeMix. Modern purebred dogs exhibited markedly elevated F_ROH compared with village dogs and wolves (Fig. 2 A). Breeds such as English Setter, BH, Irish Wolfhound, and GS showed the highest mean F_ROH values, frequently exceeding 0.30–0.40. In contrast, village dogs and several indigenous or landrace populations displayed substantially lower genomic inbreeding coefficients, often below 0.10. The distribution of F_ROH within breeds was generally tight, suggesting breed-specific demographic constraints and closed breeding structures. Pairwise FST analysis revealed pronounced genetic differentiation among modern dog breeds (Fig. 2 B). Most breed pairs exhibited moderate to high levels of differentiation (FST = 0.20–0.30), with several comparisons exceeding 0.30, indicative of strong population structure. FST values within FCI-defined groups are consistently lower than those observed between groups. For instance, Maltese and Havanese, both classified in Group 9, display closer genetic affinity despite originating from distinct geographic regions. Importantly, breeds characterized by elevated genomic inbreeding (high F_ROH) tended to exhibit greater average pairwise FST values. Pearson correlation analysis confirmed a strong positive association between population-level F_ROH and mean pairwise FST (r = 0.762, p = 9.28 × 10–21), suggesting that increased genomic homozygosity is tightly coupled with enhanced between-breed genetic divergence. To further investigate the evolutionary processes underlying breed differentiation, we inferred population splits and migration events using TreeMix (Fig. 2 C). The optimal number of migration edges was determined to be m = 7 based on the Δm statistic. The resulting tree topology revealed pronounced branch elongation among modern purebred dogs, consistent with substantial genetic drift following breed formation. Although seven migration edges were inferred, most gene flow events occurred between closely related breeds or within functional clusters, rather than representing widespread admixture across major breed groups. Together with the strong positive association between F_ROH and mean pairwise FST, the TreeMix analysis supports a model in which intensive artificial breeding has simultaneously increased within-breed homozygosity and amplified between-breed genetic divergence. Genomic differentiation between working and companion dogs highlights muscle-related functional pathways To investigate the genetic basis underlying functional divergence between working and companion dog groups, we calculated FST between Group 9 (companion dogs) and four representative working dog groups (Group 1, Group 2, Group 6, and Group 8). Across the four pairwise comparisons between working dog groups and companion dogs (Group 9), the empirical thresholds defining the top 1% of the genome-wide FST distribution varied among groups, reflecting different levels of genetic differentiation (Fig. 3 A). Specifically, the top 1% FST thresholds were 0.4918 for Group 1 vs Group 9, 0.5924 for Group 2 vs Group 9, 0.6706 for Group 6 vs Group 9, and 0.4359 for Group 8 vs Group 9. Among these comparisons, Group 6 vs Group 9 exhibited the highest differentiation threshold, whereas Group 8 vs Group 9 showed the lowest. Functional enrichment analysis of candidate genes from the Group 2 vs Group 9 comparison revealed significant enrichment of biological categories associated with muscle structure and contraction. The most significantly enriched terms included actin cytoskeleton, myofibril, and contractile muscle fiber (Fig. 3 B). Additional enrichment signals were observed for myosin-related complexes, including the myosin complex, myosin filament, and myosin II complex. To further illustrate the relationships between enriched functional categories and their associated genes, we constructed a gene–concept network using cnetplot (Fig. 3 C). The network revealed that multiple candidate genes were shared among several muscle-related functional terms, forming a highly interconnected module centered on myofibril and contractile muscle fiber. Notably, several members of the myosin gene family, including MYH1, MYH2, MYH3, MYH4, MYH8, and MYL1, were linked to multiple enriched categories, highlighting their central roles in muscle structure and contraction. Additional genes such as TNNI1 and LMOD1 were also connected to key muscle-related terms, suggesting coordinated genetic differentiation in pathways related to skeletal muscle organization and contractile function. Venn analysis of the highly differentiated genes identified in the four pairwise FST comparisons revealed that, relative to Group 9, the four working dog groups each harbored both unique and shared highly differentiated genes. The number of group-specific highly differentiated genes exceeded the number of shared genes in all comparisons. A total of 42 genes were shared across all four comparisons, and these genes were located within a genomic interval on chromosome 5 (30,718,224 − 34,518,223 bp). Functional annotation based on Gene Ontology (GO) revealed that a large proportion of the 42 shared genes were associated with cellular structural components. Specifically, 30 genes were enriched in cellular anatomical entity (GO:0110165) and 10 genes were associated with protein-containing complex (GO:0032991). In addition, several genes were involved in molecular functions related to ATP-dependent activity (GO:0140657) and cytoskeletal motor activity (GO:0003774). Notably, the enrichment of ATP-dependent activity and cytoskeletal motor activity suggests potential involvement of energy metabolism and cytoskeleton-related processes. Development of a breed identification SNP panel for dog breeds To improve the efficiency of police dog breeding and genetic management, we further sought to construct a compact breed identification panel by screening highly informative ancestry-informative markers capable of discriminating among the target working dog breeds. We first calculated allele frequencies and linkage disequilibrium (LD) patterns across the eight breeds using PLINK. These statistics were subsequently used as input for AIM-generator to identify candidate ancestry-informative SNPs with high discriminatory power among breeds. Based on the ranking results, three marker panels containing 10, 100, and 200 SNPs were constructed for subsequent evaluation. PCA revealed that the 10-SNP and 100-SNP panels were insufficient to fully separate the eight dog breeds, with several samples overlapping in the PCA space when only 10 SNPs were used (Supplementary Fig. S2 ). In contrast, the 200-SNP panel achieved complete separation of all breeds along the first two principal components (Fig. 4 A). Consistently, ADMIXTURE analysis based on the 200-SNP panel identified K = 8 as the optimal clustering model, corresponding to the eight target breeds (Fig. 4 B). To further evaluate the discriminatory performance of these panels, PCA coordinates and ADMIXTURE ancestry proportions were used as input features for Random Forest and XGBoost classification models. Using 30% of the samples as an independent validation set, all three panels achieved perfect classification performance, with even the 10-SNP panel reaching 100% classification accuracy. These results indicate that although a small number of highly informative markers can already achieve perfect classification in machine-learning frameworks, the 200-SNP panel provides the most robust resolution of breed structure and therefore represents a reliable marker set for practical breed identification. Discussion In this study, we provide a comprehensive genomic characterization of eight target dog breeds, integrating genome-wide SNP data, demographic inference, and functional analyses. We show that modern dog breeds form distinct lineages radiating from an ancestral gray wolf-like gene pool, with clear patterns of genomic differentiation reinforced by intensive artificial breeding. Our analyses demonstrate that FCI-defined breed groups largely correspond to genomic trajectories, while breeds with complex admixture histories occupy intermediate positions. We further identify muscle-related pathways as key targets of selection differentiating working dogs from companion breeds, and develop a compact SNP panel capable of robustly discriminating among breeds. Mechanisms Driving Breed Differentiation Breed classification systems, such as those used by kennel clubs, are largely based on the historical working functions of breeds [ 33 ], providing an ecologically meaningful framework for investigating behavioural and genetic differentiation among dog breeds. The star-like pattern observed in PCA and PHATE embeddings reflects rapid divergence of modern breeds from a common ancestral pool. Breeds subjected to intensive artificial selection, such as German Shepherd Dogs and Belgian Malinois, occupy distal positions along the trajectories, highlighting the cumulative effects of breeding practices on genomic differentiation. In contrast, village dogs and certain indigenous populations remain near the central cluster, consistent with limited human intervention and reduced divergence [ 34 , 35 ]. F_ROH in modern breeds, combined with high pairwise FST values, indicates that closed breeding structures amplify both within-breed homozygosity and between-breed divergence [ 36 ]. The strong positive correlation between F_ROH and FST underscores that inbreeding and genetic drift are closely linked processes driving the reinforcement of breed boundaries. TreeMix analysis further revealed that most gene flow occurs within closely related breeds or within functional FCI groups, rather than across major breed categories, consistent with deliberate human management of mating practices to preserve breed identity. Thus, within-breed similarity exceeds between-breed similarity, even when dogs from different breeds share the same living environment, function, and working ability [ 37 , 38 ]. Functional Differentiation Between Working and Companion Dogs Pairwise FST analyses between working and companion dog groups revealed strong signals of divergence in genes related to muscle structure and contractile function. Enrichment analyses highlighted pathways associated with the actin cytoskeleton, myofibril, and contractile muscle fibers, with multiple myosin family genes (MYH1, MYH2, MYH3, MYH4, MYH8, MYL1) forming densely interconnected modules. Working dogs exhibited larger muscle fibers, a higher proportion of type IIa and type I fibers, and a greater number of nuclei per fiber [ 39 , 40 ]. Additional genes involved in energy metabolism and cytoskeletal motor activity suggest coordinated selection on skeletal muscle performance, likely reflecting the functional demands imposed on working breeds. Notably, a set of 42 highly differentiated genes was shared across all working-companion comparisons. While many of these genes contribute to muscle and cytoskeletal function, several are associated with health-related traits. For example, STX8 harbors multiple loci linked to hereditary hemangiosarcoma risk [ 41 ], PER1 is implicated in circadian rhythm regulation [ 42 ], and HES7 is associated with congenital vertebral malformations [ 43 ]. These observations suggest that selection for enhanced athletic performance in working breeds may inadvertently enrich alleles that predispose to certain diseases. As sporting dogs, working dogs are at a higher risk of developing orthopedic diseases than companion dogs due to breed predispositions and the increased stress related to the activities they are engaged in [ 44 – 46 ]. Breeding from within a selected population of dogs can, in a relatively short period of time, give rise to a clear change in phenotype which leads to breed development, but may also cause an increase in the occurrence of inherited diseases. SNP-Based Dog Breed Identification Accurate breed identification is essential for breeding management, particularly in working and police dog programs. Using allele frequency and LD data, we constructed compact SNP panels capable of distinguishing among eight target breeds. Although even the 10-SNP panel achieved complete separation under machine learning frameworks, PCA and ADMIXTURE analyses showed that the 200-SNP panel provides the most robust resolution of breed structure, aligning with FCI group boundaries. These panels offer cost-effective tools for pedigree validation, breed verification, and genetic monitoring in breeding programs. Importantly, precise breed identification also facilitates the detection of unintended crossbreeding and the monitoring of inbreeding levels, both of which are critical for maintaining healthy breeding populations [ 47 – 49 ]. Intensive artificial selection and closed breeding practices can increase inbreeding coefficients, leading to reduced genetic diversity and the accumulation of deleterious variants. Therefore, the application of efficient SNP panels not only supports breed authentication but also contributes to the management of genetic diversity and the long-term sustainability of working dog populations. Conclusions This study reveals how selective breeding has shaped the genetics of working and companion dogs. Understanding these genetic differences can inform and improve breeding programs, helping police, military, and service dogs perform better while maintaining healthy populations. The identification of a small panel of genetic markers allows rapid and reliable breed identification, which can minimize misclassification and support responsible dog ownership. Overall, our findings contribute to animal welfare, enhance the utility of working dogs in society, and provide tools for better management of dog breeds worldwide. Declarations Author Contributions Rongxing Wei and Jiaqian Le contributed equally to this work. Conceptualization, Project Administration, and Writing - Original Draft Preparation, Rongxing Wei and Jiaqian Le. Resources and Writing - Review and Editing, Chao Liu and Weian Du. Supervision and Funding Acquisition, Ling Chen and Weian Du. Methodology, Linying Ye. Software, Rongxing Wei. Validation, Jiarong Chang. Formal Analysis, Linyuan Guo. Investigation, Chang Su and Mingyue Zhao. Data Curation, Weibin Wu. All authors have read and agreed to the published version of the manuscript. Funding information This work was supported by the Jiangxi Provincial Program for Selected Candidates to Lead Key Research Projects (No. 20223BBG71020). Ethics approval This study was approved by the Ethics Committee of the Nanchang Police Dog Base (No. GNJ20240103). All DNA samples were obtained from dogs with the owners’ informed consent, in compliance with institutional and national animal welfare guidelines. Data and model availability statement All genomic data generated in this study have been deposited in the OMIX database in accordance with the BMC Genomics data policy (Accession number: OMIX016186). The dataset is currently undergoing curator review and is expected to be publicly released shortly. Acknowledgements We are grateful to all colleagues and volunteers who contributed support for this study. Declaration of interest The authors declare no conflict of interest. References Savolainen P, Zhang YP, Luo J, Lundeberg J, Leitner T. Genetic evidence for an East Asian origin of domestic dogs. Science. 2002;298(5598):1610–3. Thalmann O, Shapiro B, Cui P, Schuenemann VJ, Sawyer SK, Greenfield DL, Germonpré MB, Sablin MV, López-Giráldez F, Domingo-Roura X, et al. Complete mitochondrial genomes of ancient canids suggest a European origin of domestic dogs. Science. 2013;342(6160):871–4. Frantz LA, Mullin VE, Pionnier-Capitan M, Lebrasseur O, Ollivier M, Perri A, Linderholm A, Mattiangeli V, Teasdale MD, Dimopoulos EA, et al. Genomic and archaeological evidence suggest a dual origin of domestic dogs. Science. 2016;352(6290):1228–31. Wang GD, Zhai W, Yang HC, Wang L, Zhong L, Liu YH, Fan RX, Yin TT, Zhu CL, Poyarkov AD, et al. Out of southern East Asia: the natural history of domestic dogs across the world. Cell Res. 2016;26(1):21–33. Zhou QJ, Liu X, Zhang L, Wang R, Yin T, Li X, Li G, He Y, Ding Z, Ma P, et al. A single-nucleus transcriptomic atlas of the dog hippocampus reveals the potential relationship between specific cell types and domestication. Natl Sci Rev. 2022;9(11):nwac147. Sutter NB, Ostrander EA. Dog star rising: the canine genetic system. Nat Rev Genet. 2004;5(12):900–10. Rimbault M, Beale HC, Schoenebeck JJ, Hoopes BC, Allen JJ, Kilroy-Glynn P, Wayne RK, Sutter NB, Ostrander EA. Derived variants at six genes explain nearly half of size reduction in dog breeds. Genome Res. 2013;23(12):1985–95. Plassais J, Kim J, Davis BW, Karyadi DM, Hogan AN, Harris AC, Decker B, Parker HG, Ostrander EA. Whole genome sequencing of canids reveals genomic regions under selection and variants influencing morphology. Nat Commun. 2019;10(1):1489. Rudolph TE, Ramey AR, Cullen JN, Friedenberg SG, Clark LA. Identification of genetic variants associated with ear length in drop-eared dogs. Sci Rep. 2025;16(1):3105. Shearin AL, Hedan B, Cadieu E, Erich SA, Schmidt EV, Faden DL, Cullen J, Abadie J, Kwon EM, Gröne A, et al. The MTAP-CDKN2A Locus Confers Susceptibility to a Naturally Occurring Canine Cancer. Cancer Epidemiol Biomarkers Prev. 2012;21(7):1019–27. Karyadi DM, Karlins E, Decker B, vonHoldt BM, Carpintero-Ramirez G, Parker HG, Wayne RK, Ostrander EA. A Copy Number Variant at the KITLG Locus Likely Confers Risk for Canine Squamous Cell Carcinoma of the Digit. PLoS Genet. 2013;9(3):e1003409. Karlsson EK, Sigurdsson S, Ivansson E, Thomas R, Elvers I, Wright J, Howald C, Tonomura N, Perloski M, Swofford R, et al. Genome-wide analyses implicate 33 loci in heritable dog osteosarcoma, including regulatory variants near CDKN2A/B. Genome Biol. 2013;14(12):R132. Schiffman JD, Breen M. Comparative oncology: what dogs and other species can teach us about humans with cancer. Philosophical Trans Royal Soc B: Biol Sci 2015, 370(1673). Koch Z, Graves JL, Annan S, Frankino PA, Nelson M, McMahon JE, Tovar AP, Chen HC, Halioua-Haubold CL, Peloquin M. Multi-omic analysis of canine aging uncovers conserved aging pathways. Geroscience 2025. Das S, Schlemmer SN, Idate R, Lana SE, Regan DP, Thamm DH, Duval DL. Comparative genomic and clinicopathological analysis uncovers contrasting molecular profiles of canine and human thyroid carcinomas. Commun Biol. 2025;9(1):4. Zhang S-J, Scarsbrook L, Li H, Carmagnini A, Charlton S, Feuerborn T, Boeskorov G, Chen G, Deom J-M, Dimopoulos EA, et al. Genomic evidence for the Holocene codispersal of dogs and humans across Eastern Eurasia. Science. 2025;390(6774):735–40. Boyko AR, Quignon P, Li L, Schoenebeck JJ, Degenhardt JD, Lohmueller KE, Zhao K, Brisbin A, Parker HG, vonHoldt BM, et al. A Simple Genetic Architecture Underlies Morphological Variation in Dogs. PLoS Biol. 2010;8(8):e1000451. vonHoldt BM, Pollinger JP, Lohmueller KE, Han E, Parker HG, Quignon P, Degenhardt JD, Boyko AR, Earl DA, Auton A, et al. Genome-wide SNP and haplotype analyses reveal a rich history underlying dog domestication. Nature. 2010;464(7290):898–902. Ostrander EA, Wayne RK, Freedman AH, Davis BW. Demographic history, selection and functional diversity of the canine genome. Nat Rev Genet. 2017;18(12):705–20. Parker HG, Dreger DL, Rimbault M, Davis BW, Mullen AB, Carpintero-Ramirez G, Ostrander EA. Genomic Analyses Reveal the Influence of Geographic Origin, Migration, and Hybridization on Modern Dog Breed Development. Cell Rep. 2017;19(4):697–708. Dutrow EV, Serpell JA, Ostrander EA. Domestic dog lineages reveal genetic drivers of behavioral diversification. Cell. 2022;185(25):4737–e47554718. Shannon LM, Boyko RH, Castelhano M, Corey E, Hayward JJ, McLean C, White ME, Abi Said M, Anita BA, Bondjengo NI et al. Genetic structure in village dogs reveals a Central Asian domestication origin. Proceedings of the National Academy of Sciences 2015, 112(44):13639–13644. Patterson N, Price AL, Reich D. Population Structure and Eigenanalysis. PLoS Genet. 2006;2(12):e190. Moon KR, van Dijk D, Wang Z, Gigante S, Burkhardt DB, Chen WS, Yim K, Elzen Avd, Hirn MJ, Coifman RR, et al. Visualizing structure and transitions in high-dimensional biological data. Nat Biotechnol. 2019;37(12):1482–92. Street K, Risso D, Fletcher RB, Das D, Ngai J, Yosef N, Purdom E, Dudoit S. Slingshot: cell lineage and pseudotime inference for single-cell transcriptomics. BMC Genomics. 2018;19(1):477. Chang CC, Chow CC, Tellier LCAM, Vattikuti S, Purcell SM, Lee JJ. Second-generation PLINK: rising to the challenge of larger and richer datasets. GigaScience 2015, 4(1):s13742-13015-10047-13748. Alexander DH, Novembre J, Lange K. Fast model-based estimation of ancestry in unrelated individuals. Genome Res. 2009;19(9):1655–64. Chen S, Lei C, Zhao X, Pan Y, Lu D, Xu S. AncestryPainter 2.0: Visualizing Ancestry Composition and Admixture History Graph. Genome Biol Evol. 2024;16(11):evae249. Weir BS, Cockerham CC, ESTIMATING F-STATISTICS FOR THE ANALYSIS OF POPULATION STRUCTURE. Evolution. 1984;38(6):1358–70. Pickrell JK, Pritchard JK. Inference of population splits and mixtures from genome-wide allele frequency data. PLoS Genet. 2012;8(11):e1002967. Yang Q, Chen J, Nie S, Liu C, Deng H, He G. Fine-scale biogeographical ancestry inference in Southeast and East Asians via high-efficiency markers and machine learning approaches. Front Ecol Evol 2025, Volume 13–2025. Wang GD, Cheng LG, Fan RX, Irwin DM, Tang SS, Peng JG, Zhang YP. Signature of balancing selection at the MC1R gene in Kunming dog populations. PLoS ONE. 2013;8(2):e55469. Pongrácz P, Dobos P. Behavioural differences and similarities between dog breeds: proposing an ecologically valid approach for canine behavioural research. Biol Rev. 2025;100(1):68–84. Larson G, Karlsson EK, Perri A, Webster MT, Ho SY, Peters J, Stahl PW, Piper PJ, Lingaas F, Fredholm M, et al. Rethinking dog domestication by integrating genetics, archeology, and biogeography. Proc Natl Acad Sci U S A. 2012;109(23):8878–83. Parker HG, Kim LV, Sutter NB, Carlson S, Lorentzen TD, Malek TB, Johnson GS, DeFrance HB, Ostrander EA, Kruglyak L. Genetic Structure of the Purebred Domestic Dog. Science. 2004;304(5674):1160–4. Letko A, Minor KM, Jagannathan V, Seefried FR, Mickelson JR, Oliehoek P, Drögemüller C. Genomic diversity and population structure of the Leonberger dog breed. Genet Selection Evol. 2020;52(1):61. Bigi D, Marelli SP, Liotta L, Frattini S, Talenti A, Pagnacco G, Polli M, Crepaldi P. Investigating the population structure and genetic differentiation of livestock guard dog breeds. Animal. 2018;12(10):2009–16. Suárez NM, Betancor E, Fregel R, Pestano J. Genetic characterization, at the mitochondrial and nuclear DNA levels, of five Canary Island dog breeds. Anim Genet. 2013;44(4):432–41. Osiak-Wicha C, Kras K, Arciszewski MB. Comparative Analysis of Muscle Fibers in Selected Muscles of Working and Companion Dog Breeds. Animals (Basel) 2024, 14(24). Acevedo LM, Rivero JL. New insights into skeletal muscle fibre types in the dog with particular focus towards hybrid myosin phenotypes. Cell Tissue Res. 2006;323(2):283–303. Tonomura N, Elvers I, Thomas R, Megquier K, Turner-Maier J, Howald C, Sarver AL, Swofford R, Frantz AM, Ito D, et al. Genome-wide association study identifies shared risk loci common to two malignancies in golden retrievers. PLoS Genet. 2015;11(2):e1004922. Ohmori K, Nishikawa S, Oku K, Oida K, Amagai Y, Kajiwara N, Jung K, Matsuda A, Tanaka A, Matsuda H. Circadian rhythms and the effect of glucocorticoids on expression of the clock gene period1 in canine peripheral blood mononuclear cells. Vet J. 2013;196(3):402–7. Willet CE, Makara M, Reppas G, Tsoukalas G, Malik R, Haase B, Wade CM. Canine disorder mirrors human disease: exonic deletion in HES7 causes autosomal recessive spondylocostal dysostosis in miniature Schnauzer dogs. PLoS ONE. 2015;10(2):e0117055. Alves JC, Dos Santos AM, Fernandes ÂD. Evaluation of the effect of mesotherapy in the management of back pain in police working dog. Vet Anaesth Analg. 2018;45(1):123–8. Evans RI, Herbold JR, Bradshaw BS, Moore GE. Causes for discharge of military working dogs from service: 268 cases (2000–2004). J Am Vet Med Assoc. 2007;231(8):1215–20. Worth AJ, Sandford M, Gibson B, Stratton R, Erceg V, Bridges J, Jones B. Causes of loss or retirement from active duty for New Zealand police German shepherd dogs. Anim Welf. 2013;22(2):167–74. Leroy G. Genetic diversity, inbreeding and breeding practices in dogs: results from pedigree analyses. Vet J. 2011;189(2):177–82. Kropatsch R, Melis C, Stronen AV, Jensen H, Epplen JT. Molecular Genetics of Sex Identification, Breed Ancestry and Polydactyly in the Norwegian Lundehund Breed. J Hered. 2015;106(4):403–6. Mellanby RJ, Ogden R, Clements DN, French AT, Gow AG, Powell R, Corcoran B, Schoeman JP, Summers KM. Population structure and genetic heterogeneity in popular dog breeds in the UK. Vet J. 2013;196(1):92–7. Additional Declarations No competing interests reported. Supplementary Files Supplementarymaterial.docx Cite Share Download PDF Status: Under Review Version 1 posted Reviews received at journal 13 May, 2026 Reviewers agreed at journal 28 Apr, 2026 Reviewers agreed at journal 28 Apr, 2026 Reviews received at journal 28 Apr, 2026 Reviewers agreed at journal 25 Apr, 2026 Reviewers agreed at journal 25 Apr, 2026 Reviewers agreed at journal 23 Apr, 2026 Reviewers invited by journal 23 Apr, 2026 Editor assigned by journal 23 Apr, 2026 Editor invited by journal 13 Apr, 2026 Submission checks completed at journal 13 Apr, 2026 First submitted to journal 12 Apr, 2026 You are reading this latest preprint version Research Square lets you share your work early, gain feedback from the community, and start making changes to your manuscript prior to peer review in a journal. As a division of Research Square Company, we’re committed to making research communication faster, fairer, and more useful. We do this by developing innovative software and high quality services for the global research community. Our growing team is made up of researchers and industry professionals working together to solve the most critical problems facing scientific publishing. Also discoverable on Platform About Our Team In Review Editorial Policies Advisory Board Help Center Resources Author Services Accessibility API Access RSS feed Manage Cookie Preferences © Research Square 2026 | ISSN 2693-5015 (online) Privacy Policy Terms of Service Do Not Sell My Personal Information {"props":{"pageProps":{"initialData":{"identity":"rs-9303752","acceptedTermsAndConditions":true,"allowDirectSubmit":false,"archivedVersions":[],"articleType":"Research Article","associatedPublications":[],"authors":[{"id":633529908,"identity":"d77aeff7-a48a-4738-a87e-c71454d86ecb","order_by":0,"name":"Rongxing Wei","email":"","orcid":"","institution":"Guangzhou Key Laboratory of Forensic Multi-Omics for Precision Identification, School of Forensic Medicine, Southern Medical University, Guangdong 510515","correspondingAuthor":false,"prefix":"","firstName":"Rongxing","middleName":"","lastName":"Wei","suffix":""},{"id":633529909,"identity":"d4e72749-7a96-40ce-8940-4effe4468dd0","order_by":1,"name":"jiaqian Le","email":"","orcid":"","institution":"Guangzhou Key Laboratory of Forensic Multi-Omics for Precision Identification, School of Forensic Medicine, Southern Medical University, Guangdong 510515","correspondingAuthor":false,"prefix":"","firstName":"jiaqian","middleName":"","lastName":"Le","suffix":""},{"id":633529910,"identity":"aac6d989-44a3-4468-bc30-324aa4a2331b","order_by":2,"name":"Linying Ye","email":"","orcid":"","institution":"Guangzhou Key Laboratory of Forensic Multi-Omics for Precision Identification, School of Forensic Medicine, Southern Medical University, Guangdong 510515","correspondingAuthor":false,"prefix":"","firstName":"Linying","middleName":"","lastName":"Ye","suffix":""},{"id":633529911,"identity":"d3ede295-07e6-4b14-be1b-a5accf50f271","order_by":3,"name":"Linyuan Guo","email":"","orcid":"","institution":"Gene Editing Technology Center of Guangdong Province, School of Medicine, Foshan University, Guangdong 528225","correspondingAuthor":false,"prefix":"","firstName":"Linyuan","middleName":"","lastName":"Guo","suffix":""},{"id":633529912,"identity":"60dc1a1b-b21f-4bd3-86ac-8ed9c74c3088","order_by":4,"name":"Chang Su","email":"","orcid":"","institution":"Guangzhou Key Laboratory of Forensic Multi-Omics for Precision Identification, School of Forensic Medicine, Southern Medical University, Guangdong 510515","correspondingAuthor":false,"prefix":"","firstName":"Chang","middleName":"","lastName":"Su","suffix":""},{"id":633529913,"identity":"5920a4f3-05b4-4d00-802a-e89a1d1dd672","order_by":5,"name":"Mingyue Zhao","email":"","orcid":"","institution":"Guangzhou Key Laboratory of Forensic Multi-Omics for Precision Identification, School of Forensic Medicine, Southern Medical University, Guangdong 510515","correspondingAuthor":false,"prefix":"","firstName":"Mingyue","middleName":"","lastName":"Zhao","suffix":""},{"id":633529914,"identity":"9656d6f0-2399-42bb-8789-404828f804be","order_by":6,"name":"Jiarong Chang","email":"","orcid":"","institution":"Guangzhou Key Laboratory of Forensic Multi-Omics for Precision Identification, School of Forensic Medicine, Southern Medical University, Guangdong 510515","correspondingAuthor":false,"prefix":"","firstName":"Jiarong","middleName":"","lastName":"Chang","suffix":""},{"id":633529915,"identity":"2ead75ac-7c25-4b5b-9534-10cb12ba1be9","order_by":7,"name":"Weibin Wu","email":"","orcid":"","institution":"Shaoguan Forensic Science Institute, Guangdong, 512029","correspondingAuthor":false,"prefix":"","firstName":"Weibin","middleName":"","lastName":"Wu","suffix":""},{"id":633529916,"identity":"4f578699-24d7-4996-8d63-0311c0168467","order_by":8,"name":"Chao Liu","email":"","orcid":"","institution":"Guangdong Province Key Laboratory of Pharmacodynamic Constituents of TCM and New Drugs Research, College of Pharmacy, Jinan University, 510632","correspondingAuthor":false,"prefix":"","firstName":"Chao","middleName":"","lastName":"Liu","suffix":""},{"id":633529917,"identity":"b88b9a72-2fbe-4276-beec-08e20b6e8348","order_by":9,"name":"Ling Chen","email":"","orcid":"","institution":"Guangzhou Key Laboratory of Forensic Multi-Omics for Precision Identification, School of Forensic Medicine, Southern Medical University, Guangdong 510515","correspondingAuthor":false,"prefix":"","firstName":"Ling","middleName":"","lastName":"Chen","suffix":""},{"id":633529918,"identity":"d062623e-4cb8-4994-a4ff-e8ed2b765c91","order_by":10,"name":"Weian Du","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAAtElEQVRIiWNgGAWjYDCCA0CcUGEjx8befoAELQ/OpBnz8ZxJIF4L48OWw4nzJBwMiNPBdyPH8ENiQ1p6mwRDAsOPim2EtUjeyDGWSNxhk9sm3XiAsefMbcJaDG7kbpBIPJOW2yZzIIGZsY04LZt/JLYdTmeTSDAgWss2CaCWBOK1SJ55/80i4UyaYRswkA8S5Re+42nJN39U2MjLt7cffPCjgggtKOAAiepHwSgYBaNgFOACAOuIQ2GBFuxbAAAAAElFTkSuQmCC","orcid":"","institution":"Gene Editing Technology Center of Guangdong Province, School of Medicine, Foshan University, Guangdong 528225","correspondingAuthor":true,"prefix":"","firstName":"Weian","middleName":"","lastName":"Du","suffix":""}],"badges":[],"createdAt":"2026-04-02 13:24:08","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-9303752/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-9303752/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":108493871,"identity":"ff68e73b-5a82-47ba-8503-9acd41bda411","added_by":"auto","created_at":"2026-05-05 10:01:59","extension":"png","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":3589611,"visible":true,"origin":"","legend":"\u003cp\u003eDimensionality reduction and genetic ancestry of dogs. A: PCA plot showing the distribution of all dogs. B: PHATE plot delineating paths of 9 dog lineages originating from the gray wolf. Points are colored by maximum pseudotime in any lineage and plotted lines are principal curves generated with Slingshot. C: ADMIXTURE result (K=15) illustrating genetic ancestry of dogs. Bold font indicates ancestral populations.\u003c/p\u003e","description":"","filename":"Figure1.png","url":"https://assets-eu.researchsquare.com/files/rs-9303752/v1/d9dc65a0af5a4859ed82cdb9.png"},{"id":108493585,"identity":"980e689a-4fd1-438e-8ff8-342408583802","added_by":"auto","created_at":"2026-05-05 10:01:01","extension":"png","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":1332562,"visible":true,"origin":"","legend":"\u003cp\u003eGenomic homozygosity, population differentiation, and historical gene flow of dogs. A: Genome-wide inbreeding coefficients estimated from runs of homozygosity (F_ROH). B: Heatmap of pairwise genome-wide FST among dog breeds. C: Maximum-likelihood population tree inferred using TreeMix with seven migration edges (Δm = 7).\u003c/p\u003e","description":"","filename":"Figure2.png","url":"https://assets-eu.researchsquare.com/files/rs-9303752/v1/656d5f63f5fee8d30eb22247.png"},{"id":108493678,"identity":"75bf1f85-f184-43f3-b8e9-803a7e2d933b","added_by":"auto","created_at":"2026-05-05 10:01:16","extension":"png","order_by":3,"title":"Figure 3","display":"","copyAsset":false,"role":"figure","size":2004760,"visible":true,"origin":"","legend":"\u003cp\u003eGenomic differentiation and functional enrichment between working and companion dog groups. A: Manhattan plot showing genome-wide FST values between each working dog group and Group 9 (companion dogs). Concentric tracks from the inner to outer layers represent Group 1 vs Group 9, Group 2 vs Group 9, Group 6 vs Group 9, and Group 8 vs Group 9, respectively. B: Bubble plot showing Gene Ontology enrichment of candidate genes identified from the Group 2 vs Group 9 comparison. C: Gene-concept network illustrating relationships between enriched GO terms and associated genes, generated using cnetplot. D: Venn diagram showing shared and group-specific highly differentiated genes among the four pairwise FST comparisons.\u003c/p\u003e","description":"","filename":"Figure3.png","url":"https://assets-eu.researchsquare.com/files/rs-9303752/v1/198093a0db9741a9669ece8a.png"},{"id":108459990,"identity":"455d5edc-66ed-44e9-81a7-ba702602e49b","added_by":"auto","created_at":"2026-05-05 00:34:39","extension":"png","order_by":4,"title":"Figure 4","display":"","copyAsset":false,"role":"figure","size":329344,"visible":true,"origin":"","legend":"\u003cp\u003ePerformance evaluation of 200 ancestry-informative SNP panels for dog breed identification. A: PCA of eight dog breeds based on 200-SNP panel. B: ADMIXTURE analysis of the same individuals with the optimal clustering model identified at K = 8.\u003c/p\u003e","description":"","filename":"Figure4.png","url":"https://assets-eu.researchsquare.com/files/rs-9303752/v1/5837621c25d1410a123aac01.png"},{"id":108804593,"identity":"99103e0d-0bd3-4dfd-89dc-9602c903ccff","added_by":"auto","created_at":"2026-05-08 15:21:59","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":1745183,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-9303752/v1/b454a0df-ca89-4b5b-9d12-79ae92a96ccb.pdf"},{"id":108459989,"identity":"5b116155-b9bf-4c56-a7d1-bfd99402fce9","added_by":"auto","created_at":"2026-05-05 00:34:39","extension":"docx","order_by":1,"title":"","display":"","copyAsset":false,"role":"supplement","size":348860,"visible":true,"origin":"","legend":"","description":"","filename":"Supplementarymaterial.docx","url":"https://assets-eu.researchsquare.com/files/rs-9303752/v1/c0062b6a254945cf87cc8577.docx"}],"financialInterests":"No competing interests reported.","formattedTitle":"Genome-Wide Analyses of Breed Differentiation and Functional Specialization in Domestic Dogs","fulltext":[{"header":"Introduction","content":"\u003cp\u003eHuman-dog coexistence can be traced back approximately 14,000\u0026ndash;32,000 years, representing one of the earliest and longest-standing relationships between humans and animals [\u003cspan additionalcitationids=\"CR2 CR3\" citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e4\u003c/span\u003e]. During the domestication process, dogs (\u003cem\u003eCanis lupus familiaris\u003c/em\u003e) experienced extensive diversification under long-term human-mediated selection. Today, more than 450 breeds are officially recognized by major kennel organizations such as the American Kennel Club and the F\u0026eacute;d\u0026eacute;ration Cynologique Internationale [\u003cspan citationid=\"CR5\" class=\"CitationRef\"\u003e5\u003c/span\u003e]. This remarkable phenotypic diversity has made dogs an important model for comparative genomics and studies of phenotypic evolution. Genome-wide association studies have identified the genetic basis of numerous canine traits, including body size, coat color, limb length, skull morphology, and lifespan, highlighting the power of the dog model for dissecting genotype-phenotype relationships [\u003cspan additionalcitationids=\"CR7 CR8\" citationid=\"CR6\" class=\"CitationRef\"\u003e6\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR9\" class=\"CitationRef\"\u003e9\u003c/span\u003e]. Additionally, the dog model has been utilized to identify genes with translational potential for human health and biology, including both rare and common human disorders, such as autoimmune disease, neuromuscular disorders, and cancer [\u003cspan additionalcitationids=\"CR11 CR12 CR13 CR14\" citationid=\"CR10\" class=\"CitationRef\"\u003e10\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR15\" class=\"CitationRef\"\u003e15\u003c/span\u003e].\u003c/p\u003e \u003cp\u003eRecent ancient genomic studies further suggest that dogs and humans dispersed together across Eurasia, indicating a long history of co-dispersal between the two species [\u003cspan citationid=\"CR16\" class=\"CitationRef\"\u003e16\u003c/span\u003e]. In some cases, dogs possessing valuable working or physical traits were exchanged among human populations, facilitating the spread of functionally specialized lineages. However, most modern dog breeds emerged relatively recently, primarily during the mid-19th century, when deliberate reproductive isolation and intensive artificial selection led to the establishment of standardized breeds [\u003cspan additionalcitationids=\"CR18 CR19\" citationid=\"CR17\" class=\"CitationRef\"\u003e17\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR20\" class=\"CitationRef\"\u003e20\u003c/span\u003e]. These processes, often accompanied by strong population bottlenecks and closed breeding systems, generated pronounced population structure and genetic differentiation among breeds.\u003c/p\u003e \u003cp\u003eFunctional specialization has therefore been a major driver of canine diversity [\u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e8\u003c/span\u003e]. Historically, different breeds were selectively bred for specific purposes, such as hunting, guarding, herding, and companionship [\u003cspan citationid=\"CR21\" class=\"CitationRef\"\u003e21\u003c/span\u003e]. Consequently, modern dog populations can be broadly categorized into working dogs and companion dogs. However, despite their clear functional divergence, the genetic differences between working and companion dogs remain insufficiently characterized. In particular, widely used police and working breeds, including German Shepherd Dog (GS), Belgian Malinois (BM), Kunming Dog (KM), Bloodhound (BH), Labrador Retriever (LR), Rottweiler (RT), and English Springer Spaniel (SS), as well as highly popular companion breeds such as the French Bulldog (FB), provide a valuable system for investigating the genomic basis of functional specialization in domestic dogs.\u003c/p\u003e \u003cp\u003eIn this study, we aimed to investigate the genetic basis of functional differentiation among dog breeds using genome-wide SNP data. Specifically, we characterized the population structure and genetic relationships among multiple dog breeds, identified genomic regions showing strong differentiation between working and companion dogs through genome-wide FST scans, and developed compact SNP panels capable of reliably distinguishing among target breeds. These analyses provide insights into the genomic signatures associated with functional divergence in dogs and offer practical tools for breed identification and breeding management.\u003c/p\u003e"},{"header":"Material and methods","content":"\u003cdiv id=\"Sec3\" class=\"Section2\"\u003e \u003ch2\u003eSample Collection\u003c/h2\u003e \u003cp\u003eEight dog breeds were selected from the sample repository, including seven working breeds and one popular companion breed: German Shepherd Dog (GS), Belgian Malinois Shepherd Dog (BM), Kunming Dog (KM), Bloodhound (BH), Labrador Retriever (LR), Rottweiler (RT), English Springer Spaniel (SS), and French Bulldog (FB). According to the F\u0026eacute;d\u0026eacute;ration Cynologique Internationale (FCI) classification, GS and BM belong to Group 1 (Sheepdogs), RT to Group 2 (Molossoid breeds), BH to Group 6 (Scent hounds), LR and SS to Group 8 (Retrievers), and FB to Group 9 (Small Molossoid-type Dogs). KM is a Chinese native breed that is not recognized by the FCI. Samples were obtained from the Nanjing Police Dog Research Institute, Kunming Police Dog Base, and Nanchang Police Dog Base. A total of 312 individuals were included: 36 BM, 20 BH, 71 GS, 46 FB, 27 KM, 25 SS, 42 LR, and 45 RT. All sampled dogs were unrelated for at least two generations.\u003c/p\u003e \u003c/div\u003e\n\u003ch3\u003eDNA Extraction\u003c/h3\u003e\n\u003cp\u003eGenomic DNA was extracted from dried blood spots using a commercial DNA extraction kit following the manufacturer's protocol. Briefly, three 3\u0026times;3 mm blood spot punches were incubated with lysis buffer and Proteinase K until fully solubilized. DNA was then bound to the column in the presence of ethanol, followed by sequential washes to remove impurities and residual salts. Columns were air-dried to eliminate residual ethanol, and DNA was finally eluted in 20\u0026ndash;50 \u0026micro;L elution buffer for downstream analyses.\u003c/p\u003e\n\u003ch3\u003eSNP Genotyping and Quality Control\u003c/h3\u003e\n\u003cp\u003eGenomic DNA was genotyped using the Illumina Canine 230K SNP array following the manufacturer's protocol. DNA concentrations were measured and normalized to 50 ng/\u0026micro;L, with 200 ng used for the Infinium HD Assay. Genomic DNA was denatured and subjected to whole-genome amplification, yielding 2000-3000-fold increase in DNA amount. Amplified DNA was enzymatically fragmented in a controlled end-point reaction, precipitated with isopropanol, and resuspended in hybridization buffer. DNA samples were hybridized to the bead-based array overnight, allowing 50-base site-specific probes on each bead to anneal to the complementary DNA fragments. Non-specifically bound DNA was removed, followed by single-base extension and fluorescent labeling. Arrays were scanned using an Illumina scanner, and fluorescence intensities were converted to SNP genotypes using BeadStudio software. Genotyping quality was high, with a mean call rate of 96.57%. Ten individuals with call rates below 90% were excluded from further analyses. After applying additional filtering for missingness (--mind 0.01) and removing SNPs with a minor allele frequency below 0.05, a total of 290 individuals and 196,125 markers were retained for downstream analyses.\u003c/p\u003e\n\u003ch3\u003eIntegration of Modern Genomic Data\u003c/h3\u003e\n\u003cp\u003eWe further integrated publicly available genotype data from a global dataset, comprising 5,392 dogs [\u003cspan citationid=\"CR22\" class=\"CitationRef\"\u003e22\u003c/span\u003e]. Only SNPs genotyped on both arrays and passing quality control were included (69,018 SNPs). This dataset includes both breed dogs and village dogs sampled across multiple geographic regions, including Africa, America, Middle East, Australia, Central Asia, East Asia, Europe, and the Arctic. All modern breed dogs were classified according to the FCI grouping system.\u003c/p\u003e\n\u003ch3\u003ePopulation genetic analysis\u003c/h3\u003e\n\u003cdiv id=\"Sec8\" class=\"Section2\"\u003e \u003ch2\u003eDimensionality Reduction\u003c/h2\u003e \u003cp\u003ePrincipal component analysis (PCA) of unrelated modern dogs was performed using the smartpca program implemented in EIGENSOFT v3.0 [\u003cspan citationid=\"CR23\" class=\"CitationRef\"\u003e23\u003c/span\u003e]. The two-dimensional PHATE embedding was generated using the first 20 PCs as input with phate (pca_matrix, ndim\u0026thinsp;=\u0026thinsp;2, knn\u0026thinsp;=\u0026thinsp;80, gamma\u0026thinsp;=\u0026thinsp;0, decay\u0026thinsp;=\u0026thinsp;100) [\u003cspan citationid=\"CR24\" class=\"CitationRef\"\u003e24\u003c/span\u003e]. Slingshot v1.4.0 was used for identifying lineages using PHATE coordinates as input and clusters defined using kmeans (centers\u0026thinsp;=\u0026thinsp;40) [\u003cspan citationid=\"CR25\" class=\"CitationRef\"\u003e25\u003c/span\u003e].\u003c/p\u003e \u003c/div\u003e\n\u003ch3\u003eADMIXTURE Analysis\u003c/h3\u003e\n\u003cp\u003eTo estimate the ancestry of dogs, SNPs were pruned for linkage disequilibrium (LD) using PLINK v1.9 (--indep-pairwise 50 5 2) [\u003cspan citationid=\"CR26\" class=\"CitationRef\"\u003e26\u003c/span\u003e]. ADMIXTURE v1.3.0 was run on the pruned dataset for K\u0026thinsp;=\u0026thinsp;2\u0026ndash;20 with 10-fold cross-validation and 100 bootstrap replicates [\u003cspan citationid=\"CR27\" class=\"CitationRef\"\u003e27\u003c/span\u003e]. Cross-validation (CV) errors were calculated for each K to determine the optimal model, and the best-supported K was identified based on the lowest CV error. Ancestry proportions were visualized using AncestryPainterV2 [\u003cspan citationid=\"CR28\" class=\"CitationRef\"\u003e28\u003c/span\u003e].\u003c/p\u003e\n\u003ch3\u003ePairwise Fixation Index (FST) Estimation\u003c/h3\u003e\n\u003cp\u003eWe then calculated pairwise fixation indexes (FST) among geographically distinct dogs using PLINK v1.9 with the --fst flag [\u003cspan citationid=\"CR26\" class=\"CitationRef\"\u003e26\u003c/span\u003e, \u003cspan citationid=\"CR29\" class=\"CitationRef\"\u003e29\u003c/span\u003e].\u003c/p\u003e \u003cdiv id=\"Sec11\" class=\"Section2\"\u003e \u003ch2\u003eRuns of Homozygosity\u003c/h2\u003e \u003cp\u003eWe used PLINK v1.9 to identify individual-level ROHs and calculated the population average level among geographically different dogs [\u003cspan citationid=\"CR26\" class=\"CitationRef\"\u003e26\u003c/span\u003e]. For each detected ROH segment, physical length (in kilobases) was converted to megabases (Mb). For each individual, the total length of ROH segments was calculated by summing all ROH segments across autosomes. Genomic inbreeding coefficients based on ROH (F_ROH) were calculated as:\u003c/p\u003e \u003cp\u003e\u003cimg src=\"data:image/png;base64,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\" width=\"282\" height=\"70\"\u003e\u003c/p\u003e\u003cp\u003ewhere \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:\\text{L=2203.76}\\)\u003c/span\u003e\u003c/span\u003eMb corresponds to the total length of the CanFam3.1 autosomal genome.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec12\" class=\"Section2\"\u003e \u003ch2\u003eTreemix Analysis\u003c/h2\u003e \u003cp\u003ePopulation trees and historical gene flow were inferred using Treemix v1.13 [\u003cspan citationid=\"CR30\" class=\"CitationRef\"\u003e30\u003c/span\u003e]. For each number of migration edges (m\u0026thinsp;=\u0026thinsp;0\u0026ndash;7), five replicates were run with the gray wolf as outgroup, 500-SNP blocks (-k 500), bootstrap support (-bootstrap), and standard errors of migration edges (-se). Global optimization (-global) was applied and sample size correction was disabled (-noss). Runs were executed in parallel in the background. The optimal number of migration edges was determined using OptM, and trees were visualized with custom R functions and exported as PDF files.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec13\" class=\"Section2\"\u003e \u003ch2\u003eIdentification and functional characterization of highly differentiated loci\u003c/h2\u003e \u003cp\u003eEight breeds were assigned to groups following the classification of the FCI. In addition, KM was incorporated into Group 1 because of its genetic affinity to GS and their similar working functions. Group 9 represents companion dogs, whereas all other groups correspond to working dogs. We calculated pairwise genetic differentiation (FST) between Group 9 and each of the working dog groups using a sliding window approach and obtained the mean FST for each window. Genomic regions within the top 1% of the empirical FST distribution were defined as highly differentiated loci. Genes located within these candidate regions were annotated based on the Ensembl gene annotation database (CanFam3.1, release 104). To explore the potential biological functions of the candidate genes, Gene Ontology enrichment analyses were conducted using the R package clusterProfiler. In addition, gene-concept network visualization was performed using the cnetplot function to illustrate the relationships between enriched functional categories and the associated genes. Finally, candidate gene sets identified from each group comparison were compared using Venn analysis to identify genes shared across groups.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec14\" class=\"Section2\"\u003e \u003ch2\u003eIdentification of ancestry-informative SNP panels and breed classification using population structure and machine learning analyses\u003c/h2\u003e \u003cp\u003eAllele frequencies and pairwise linkage disequilibrium (LD) for each SNP were calculated in PLINK v1.9. To identify ancestry-informative markers capable of discriminating among the eight breeds, the software AIM-generator [\u003cspan citationid=\"CR31\" class=\"CitationRef\"\u003e31\u003c/span\u003e] was applied. Using this approach, three SNP panels containing 10, 100, and 200 markers were selected to evaluate their ability to distinguish among the eight dog breeds. In addition, PCA was conducted using the selected SNP sets with PLINK to visualize genetic clustering among individuals. Ancestry proportions were estimated using ADMIXTURE run assuming K\u0026thinsp;=\u0026thinsp;8 ancestral populations, corresponding to the number of breeds analyzed. To further evaluate the effectiveness of the selected SNP panels for breed classification, machine learning models (Random Forest and XGBoost) were constructed based on the results of PCA and ADMIXTURE. The dataset was randomly divided into training (70%) and validation (30%) sets, while maintaining proportional representation of each breed. Hyperparameters for each model were optimized using grid search with cross-validation to maximize classification accuracy on the training dataset. The optimized models were then applied to the validation dataset, and classification performance was evaluated based on overall accuracy and confusion matrices.\u003c/p\u003e \u003c/div\u003e"},{"header":"Results","content":"\u003cdiv id=\"Sec16\" class=\"Section2\"\u003e \u003ch2\u003eGenome-wide Dimensionality Reduction Identifies Dog Lineages Shaped by Divergent Working Roles\u003c/h2\u003e \u003cp\u003e All breed dogs in this study were categorized according to the F\u0026eacute;d\u0026eacute;ration Cynologique Internationale (FCI) classification system, with additional stratification based on their documented geographic origins to facilitate biologically meaningful comparisons.\u003c/p\u003e \u003cp\u003eTo investigate genetic relationships among individuals, we performed principal component analysis (PCA) using genome-wide SNP data. The two-dimensional projection of the first two principal components revealed robust genetic structuring across dog breeds (Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003eA). Notably, the resulting pattern exhibited a star-like configuration radiating from gray wolves, which were positioned near the center of the plot. This central placement likely reflects their ancestral status relative to domestic dogs. Most European breeds were distributed along three well-defined clines extending outward from the wolf cluster, indicating directional divergence along partially independent genetic axes. Among the eight focal dog breeds, German Shepherd Dog (GS), Belgian Malinois Shepherd Dog (BM), Kunming Dog (KM), Labrador Retriever (LR), Rottweiler (RT), and French Bulldog (FB) were distinctly separated from gray wolves along the primary axis of variation, forming a tightly grouped cluster. The KM occupied an intermediate position along one of the European-associated clines, consistent with its breeding history, which involved admixture between German Shepherd Dogs imported from the Soviet Union and indigenous village dogs from Kunming in the 1950s [\u003cspan citationid=\"CR32\" class=\"CitationRef\"\u003e32\u003c/span\u003e]. In contrast, Bloodhound (BH) and English Springer Spaniel (SS) were positioned closer to the wolf cluster, reflecting reduced genetic differentiation relative to other breeds.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003eTo capture global relationships beyond discrete breed clusters, we applied potential of heat diffusion for affinity-based transition embedding (PHATE). PHATE preserves both local and global structure in high-dimensional data and models pairwise relationships among samples as diffusion probabilities. The resulting two-dimensional embedding displayed a star-shaped configuration with gray wolves near the center and multiple trajectories radiating outward. Trajectory extremities corresponded to the most genetically differentiated breeds, whereas basal portions occupied intermediate positions relative to the central wolf cluster. These patterns illustrate the relative genetic divergence and continuity of genomic variation among dog breeds.\u003c/p\u003e \u003cp\u003eMapping breeds according to the ten FCI groups revealed that breeds from different functional or geographic categories generally followed distinct trajectories within the PHATE manifold (Supplementary Fig. \u003cspan refid=\"MOESM1\" class=\"InternalRef\"\u003eS1\u003c/span\u003e). For example, Group 10 breeds aligned along one trajectory, Group 5 along another, and freely breeding village dogs clustered near the center, consistent with lower differentiation. To formally characterize these patterns, we applied pseudotemporal reconstruction, which orders clusters of individuals along branching trajectories in high-dimensional space. Clusters were defined using k-means clustering, with gray wolves serving as a central reference. In this unsupervised framework, nine major lineages were identified, whereas breeds in Group 4 and Group 6 remained near the central region, reflecting their relatively low genetic differentiation compared to other modern breeds (Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003eB).\u003c/p\u003e \u003cp\u003eTo further examine ancestral composition, we performed ADMIXTURE analysis at K\u0026thinsp;=\u0026thinsp;15, which minimized cross-validation error. The results indicated that GS, RT, LR, and FB each possessed predominantly distinct ancestral components (Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003eC). In contrast, KM, BH, BM, and ESS displayed more admixed patterns. Notably, the Kunming Dog genome comprised over 50 percent ancestry related to the German Shepherd Dog, whereas Bloodhounds, Belgian Shepherd Dogs, and English Springer Spaniels contained varying proportions of ancestry associated with the English Springer Spaniel ancestral component.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec17\" class=\"Section2\"\u003e \u003ch2\u003eGenomic Inbreeding and Genetic Drift Reinforce Breed Differentiation\u003c/h2\u003e \u003cp\u003eTo investigate the demographic and evolutionary processes underlying the lineage structure identified above, we examined patterns of runs of homozygosity (ROH) within breeds, genome-wide genetic differentiation among breeds, and historical gene flow using TreeMix. Modern purebred dogs exhibited markedly elevated F_ROH compared with village dogs and wolves (Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003eA). Breeds such as English Setter, BH, Irish Wolfhound, and GS showed the highest mean F_ROH values, frequently exceeding 0.30\u0026ndash;0.40. In contrast, village dogs and several indigenous or landrace populations displayed substantially lower genomic inbreeding coefficients, often below 0.10. The distribution of F_ROH within breeds was generally tight, suggesting breed-specific demographic constraints and closed breeding structures. Pairwise FST analysis revealed pronounced genetic differentiation among modern dog breeds (Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003eB). Most breed pairs exhibited moderate to high levels of differentiation (FST\u0026thinsp;=\u0026thinsp;0.20\u0026ndash;0.30), with several comparisons exceeding 0.30, indicative of strong population structure. FST values within FCI-defined groups are consistently lower than those observed between groups. For instance, Maltese and Havanese, both classified in Group 9, display closer genetic affinity despite originating from distinct geographic regions. Importantly, breeds characterized by elevated genomic inbreeding (high F_ROH) tended to exhibit greater average pairwise FST values. Pearson correlation analysis confirmed a strong positive association between population-level F_ROH and mean pairwise FST (r\u0026thinsp;=\u0026thinsp;0.762, p\u0026thinsp;=\u0026thinsp;9.28 \u0026times; 10\u0026ndash;21), suggesting that increased genomic homozygosity is tightly coupled with enhanced between-breed genetic divergence. To further investigate the evolutionary processes underlying breed differentiation, we inferred population splits and migration events using TreeMix (Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003eC). The optimal number of migration edges was determined to be m\u0026thinsp;=\u0026thinsp;7 based on the Δm statistic. The resulting tree topology revealed pronounced branch elongation among modern purebred dogs, consistent with substantial genetic drift following breed formation. Although seven migration edges were inferred, most gene flow events occurred between closely related breeds or within functional clusters, rather than representing widespread admixture across major breed groups. Together with the strong positive association between F_ROH and mean pairwise FST, the TreeMix analysis supports a model in which intensive artificial breeding has simultaneously increased within-breed homozygosity and amplified between-breed genetic divergence.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec18\" class=\"Section2\"\u003e \u003ch2\u003eGenomic differentiation between working and companion dogs highlights muscle-related functional pathways\u003c/h2\u003e \u003cp\u003eTo investigate the genetic basis underlying functional divergence between working and companion dog groups, we calculated FST between Group 9 (companion dogs) and four representative working dog groups (Group 1, Group 2, Group 6, and Group 8). Across the four pairwise comparisons between working dog groups and companion dogs (Group 9), the empirical thresholds defining the top 1% of the genome-wide FST distribution varied among groups, reflecting different levels of genetic differentiation (Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003eA). Specifically, the top 1% FST thresholds were 0.4918 for Group 1 vs Group 9, 0.5924 for Group 2 vs Group 9, 0.6706 for Group 6 vs Group 9, and 0.4359 for Group 8 vs Group 9. Among these comparisons, Group 6 vs Group 9 exhibited the highest differentiation threshold, whereas Group 8 vs Group 9 showed the lowest. Functional enrichment analysis of candidate genes from the Group 2 vs Group 9 comparison revealed significant enrichment of biological categories associated with muscle structure and contraction. The most significantly enriched terms included actin cytoskeleton, myofibril, and contractile muscle fiber (Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003eB). Additional enrichment signals were observed for myosin-related complexes, including the myosin complex, myosin filament, and myosin II complex. To further illustrate the relationships between enriched functional categories and their associated genes, we constructed a gene\u0026ndash;concept network using cnetplot (Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003eC). The network revealed that multiple candidate genes were shared among several muscle-related functional terms, forming a highly interconnected module centered on myofibril and contractile muscle fiber. Notably, several members of the myosin gene family, including MYH1, MYH2, MYH3, MYH4, MYH8, and MYL1, were linked to multiple enriched categories, highlighting their central roles in muscle structure and contraction. Additional genes such as TNNI1 and LMOD1 were also connected to key muscle-related terms, suggesting coordinated genetic differentiation in pathways related to skeletal muscle organization and contractile function. Venn analysis of the highly differentiated genes identified in the four pairwise FST comparisons revealed that, relative to Group 9, the four working dog groups each harbored both unique and shared highly differentiated genes. The number of group-specific highly differentiated genes exceeded the number of shared genes in all comparisons. A total of 42 genes were shared across all four comparisons, and these genes were located within a genomic interval on chromosome 5 (30,718,224\u0026thinsp;\u0026minus;\u0026thinsp;34,518,223 bp). Functional annotation based on Gene Ontology (GO) revealed that a large proportion of the 42 shared genes were associated with cellular structural components. Specifically, 30 genes were enriched in cellular anatomical entity (GO:0110165) and 10 genes were associated with protein-containing complex (GO:0032991). In addition, several genes were involved in molecular functions related to ATP-dependent activity (GO:0140657) and cytoskeletal motor activity (GO:0003774). Notably, the enrichment of ATP-dependent activity and cytoskeletal motor activity suggests potential involvement of energy metabolism and cytoskeleton-related processes.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec19\" class=\"Section2\"\u003e \u003ch2\u003eDevelopment of a breed identification SNP panel for dog breeds\u003c/h2\u003e \u003cp\u003eTo improve the efficiency of police dog breeding and genetic management, we further sought to construct a compact breed identification panel by screening highly informative ancestry-informative markers capable of discriminating among the target working dog breeds. We first calculated allele frequencies and linkage disequilibrium (LD) patterns across the eight breeds using PLINK. These statistics were subsequently used as input for AIM-generator to identify candidate ancestry-informative SNPs with high discriminatory power among breeds. Based on the ranking results, three marker panels containing 10, 100, and 200 SNPs were constructed for subsequent evaluation. PCA revealed that the 10-SNP and 100-SNP panels were insufficient to fully separate the eight dog breeds, with several samples overlapping in the PCA space when only 10 SNPs were used (Supplementary Fig. \u003cspan refid=\"MOESM2\" class=\"InternalRef\"\u003eS2\u003c/span\u003e). In contrast, the 200-SNP panel achieved complete separation of all breeds along the first two principal components (Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003eA). Consistently, ADMIXTURE analysis based on the 200-SNP panel identified K\u0026thinsp;=\u0026thinsp;8 as the optimal clustering model, corresponding to the eight target breeds (Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003eB). To further evaluate the discriminatory performance of these panels, PCA coordinates and ADMIXTURE ancestry proportions were used as input features for Random Forest and XGBoost classification models. Using 30% of the samples as an independent validation set, all three panels achieved perfect classification performance, with even the 10-SNP panel reaching 100% classification accuracy. These results indicate that although a small number of highly informative markers can already achieve perfect classification in machine-learning frameworks, the 200-SNP panel provides the most robust resolution of breed structure and therefore represents a reliable marker set for practical breed identification.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003c/div\u003e"},{"header":"Discussion","content":"\u003cp\u003eIn this study, we provide a comprehensive genomic characterization of eight target dog breeds, integrating genome-wide SNP data, demographic inference, and functional analyses. We show that modern dog breeds form distinct lineages radiating from an ancestral gray wolf-like gene pool, with clear patterns of genomic differentiation reinforced by intensive artificial breeding. Our analyses demonstrate that FCI-defined breed groups largely correspond to genomic trajectories, while breeds with complex admixture histories occupy intermediate positions. We further identify muscle-related pathways as key targets of selection differentiating working dogs from companion breeds, and develop a compact SNP panel capable of robustly discriminating among breeds.\u003c/p\u003e \u003cdiv id=\"Sec21\" class=\"Section2\"\u003e \u003ch2\u003eMechanisms Driving Breed Differentiation\u003c/h2\u003e \u003cp\u003eBreed classification systems, such as those used by kennel clubs, are largely based on the historical working functions of breeds [\u003cspan citationid=\"CR33\" class=\"CitationRef\"\u003e33\u003c/span\u003e], providing an ecologically meaningful framework for investigating behavioural and genetic differentiation among dog breeds. The star-like pattern observed in PCA and PHATE embeddings reflects rapid divergence of modern breeds from a common ancestral pool. Breeds subjected to intensive artificial selection, such as German Shepherd Dogs and Belgian Malinois, occupy distal positions along the trajectories, highlighting the cumulative effects of breeding practices on genomic differentiation. In contrast, village dogs and certain indigenous populations remain near the central cluster, consistent with limited human intervention and reduced divergence [\u003cspan citationid=\"CR34\" class=\"CitationRef\"\u003e34\u003c/span\u003e, \u003cspan citationid=\"CR35\" class=\"CitationRef\"\u003e35\u003c/span\u003e]. F_ROH in modern breeds, combined with high pairwise FST values, indicates that closed breeding structures amplify both within-breed homozygosity and between-breed divergence [\u003cspan citationid=\"CR36\" class=\"CitationRef\"\u003e36\u003c/span\u003e]. The strong positive correlation between F_ROH and FST underscores that inbreeding and genetic drift are closely linked processes driving the reinforcement of breed boundaries. TreeMix analysis further revealed that most gene flow occurs within closely related breeds or within functional FCI groups, rather than across major breed categories, consistent with deliberate human management of mating practices to preserve breed identity. Thus, within-breed similarity exceeds between-breed similarity, even when dogs from different breeds share the same living environment, function, and working ability [\u003cspan citationid=\"CR37\" class=\"CitationRef\"\u003e37\u003c/span\u003e, \u003cspan citationid=\"CR38\" class=\"CitationRef\"\u003e38\u003c/span\u003e].\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec22\" class=\"Section2\"\u003e \u003ch2\u003eFunctional Differentiation Between Working and Companion Dogs\u003c/h2\u003e \u003cp\u003ePairwise FST analyses between working and companion dog groups revealed strong signals of divergence in genes related to muscle structure and contractile function. Enrichment analyses highlighted pathways associated with the actin cytoskeleton, myofibril, and contractile muscle fibers, with multiple myosin family genes (MYH1, MYH2, MYH3, MYH4, MYH8, MYL1) forming densely interconnected modules. Working dogs exhibited larger muscle fibers, a higher proportion of type IIa and type I fibers, and a greater number of nuclei per fiber [\u003cspan citationid=\"CR39\" class=\"CitationRef\"\u003e39\u003c/span\u003e, \u003cspan citationid=\"CR40\" class=\"CitationRef\"\u003e40\u003c/span\u003e]. Additional genes involved in energy metabolism and cytoskeletal motor activity suggest coordinated selection on skeletal muscle performance, likely reflecting the functional demands imposed on working breeds. Notably, a set of 42 highly differentiated genes was shared across all working-companion comparisons. While many of these genes contribute to muscle and cytoskeletal function, several are associated with health-related traits. For example, STX8 harbors multiple loci linked to hereditary hemangiosarcoma risk [\u003cspan citationid=\"CR41\" class=\"CitationRef\"\u003e41\u003c/span\u003e], PER1 is implicated in circadian rhythm regulation [\u003cspan citationid=\"CR42\" class=\"CitationRef\"\u003e42\u003c/span\u003e], and HES7 is associated with congenital vertebral malformations [\u003cspan citationid=\"CR43\" class=\"CitationRef\"\u003e43\u003c/span\u003e]. These observations suggest that selection for enhanced athletic performance in working breeds may inadvertently enrich alleles that predispose to certain diseases. As sporting dogs, working dogs are at a higher risk of developing orthopedic diseases than companion dogs due to breed predispositions and the increased stress related to the activities they are engaged in [\u003cspan additionalcitationids=\"CR45\" citationid=\"CR44\" class=\"CitationRef\"\u003e44\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR46\" class=\"CitationRef\"\u003e46\u003c/span\u003e]. Breeding from within a selected population of dogs can, in a relatively short period of time, give rise to a clear change in phenotype which leads to breed development, but may also cause an increase in the occurrence of inherited diseases.\u003c/p\u003e \u003cdiv id=\"Sec23\" class=\"Section3\"\u003e \u003ch2\u003eSNP-Based Dog Breed Identification\u003c/h2\u003e \u003cp\u003eAccurate breed identification is essential for breeding management, particularly in working and police dog programs. Using allele frequency and LD data, we constructed compact SNP panels capable of distinguishing among eight target breeds. Although even the 10-SNP panel achieved complete separation under machine learning frameworks, PCA and ADMIXTURE analyses showed that the 200-SNP panel provides the most robust resolution of breed structure, aligning with FCI group boundaries. These panels offer cost-effective tools for pedigree validation, breed verification, and genetic monitoring in breeding programs. Importantly, precise breed identification also facilitates the detection of unintended crossbreeding and the monitoring of inbreeding levels, both of which are critical for maintaining healthy breeding populations [\u003cspan additionalcitationids=\"CR48\" citationid=\"CR47\" class=\"CitationRef\"\u003e47\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR49\" class=\"CitationRef\"\u003e49\u003c/span\u003e]. Intensive artificial selection and closed breeding practices can increase inbreeding coefficients, leading to reduced genetic diversity and the accumulation of deleterious variants. Therefore, the application of efficient SNP panels not only supports breed authentication but also contributes to the management of genetic diversity and the long-term sustainability of working dog populations.\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e"},{"header":"Conclusions","content":"\u003cp\u003eThis study reveals how selective breeding has shaped the genetics of working and companion dogs. Understanding these genetic differences can inform and improve breeding programs, helping police, military, and service dogs perform better while maintaining healthy populations. The identification of a small panel of genetic markers allows rapid and reliable breed identification, which can minimize misclassification and support responsible dog ownership. Overall, our findings contribute to animal welfare, enhance the utility of working dogs in society, and provide tools for better management of dog breeds worldwide.\u003c/p\u003e"},{"header":"Declarations","content":"\u003cp\u003e\u003cstrong\u003eAuthor Contributions\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eRongxing Wei and Jiaqian Le contributed equally to this work. Conceptualization, Project Administration, and Writing - Original Draft Preparation, Rongxing Wei and Jiaqian Le. Resources and Writing - Review and Editing, Chao Liu and Weian Du. Supervision and Funding Acquisition, Ling Chen and Weian Du. Methodology, Linying Ye. Software, Rongxing Wei. Validation, Jiarong Chang. Formal Analysis, Linyuan Guo. Investigation, Chang Su and Mingyue Zhao. Data Curation, Weibin Wu. All authors have read and agreed to the published version of the manuscript.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eFunding information\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThis work was supported by the Jiangxi Provincial Program for Selected Candidates to Lead Key Research Projects (No. 20223BBG71020).\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eEthics approval\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThis study was approved by the Ethics Committee of the Nanchang Police Dog Base (No. GNJ20240103). All DNA samples were obtained from dogs with the owners\u0026rsquo; informed consent, in compliance with institutional and national animal welfare guidelines.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eData and model availability statement\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eAll genomic data generated in this study have been deposited in the OMIX database in accordance with the BMC Genomics data policy (Accession number: OMIX016186). The dataset is currently undergoing curator review and is expected to be publicly released shortly.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAcknowledgements\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eWe are grateful to all colleagues and volunteers who contributed support for this study.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eDeclaration of interest\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe authors declare no conflict of interest.\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\u003cli\u003e\u003cspan\u003eSavolainen P, Zhang YP, Luo J, Lundeberg J, Leitner T. Genetic evidence for an East Asian origin of domestic dogs. Science. 2002;298(5598):1610\u0026ndash;3.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eThalmann O, Shapiro B, Cui P, Schuenemann VJ, Sawyer SK, Greenfield DL, Germonpr\u0026eacute; MB, Sablin MV, L\u0026oacute;pez-Gir\u0026aacute;ldez F, Domingo-Roura X, et al. Complete mitochondrial genomes of ancient canids suggest a European origin of domestic dogs. Science. 2013;342(6160):871\u0026ndash;4.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eFrantz LA, Mullin VE, Pionnier-Capitan M, Lebrasseur O, Ollivier M, Perri A, Linderholm A, Mattiangeli V, Teasdale MD, Dimopoulos EA, et al. Genomic and archaeological evidence suggest a dual origin of domestic dogs. Science. 2016;352(6290):1228\u0026ndash;31.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eWang GD, Zhai W, Yang HC, Wang L, Zhong L, Liu YH, Fan RX, Yin TT, Zhu CL, Poyarkov AD, et al. Out of southern East Asia: the natural history of domestic dogs across the world. Cell Res. 2016;26(1):21\u0026ndash;33.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eZhou QJ, Liu X, Zhang L, Wang R, Yin T, Li X, Li G, He Y, Ding Z, Ma P, et al. A single-nucleus transcriptomic atlas of the dog hippocampus reveals the potential relationship between specific cell types and domestication. Natl Sci Rev. 2022;9(11):nwac147.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eSutter NB, Ostrander EA. Dog star rising: the canine genetic system. Nat Rev Genet. 2004;5(12):900\u0026ndash;10.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eRimbault M, Beale HC, Schoenebeck JJ, Hoopes BC, Allen JJ, Kilroy-Glynn P, Wayne RK, Sutter NB, Ostrander EA. Derived variants at six genes explain nearly half of size reduction in dog breeds. Genome Res. 2013;23(12):1985\u0026ndash;95.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003ePlassais J, Kim J, Davis BW, Karyadi DM, Hogan AN, Harris AC, Decker B, Parker HG, Ostrander EA. Whole genome sequencing of canids reveals genomic regions under selection and variants influencing morphology. Nat Commun. 2019;10(1):1489.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eRudolph TE, Ramey AR, Cullen JN, Friedenberg SG, Clark LA. Identification of genetic variants associated with ear length in drop-eared dogs. Sci Rep. 2025;16(1):3105.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eShearin AL, Hedan B, Cadieu E, Erich SA, Schmidt EV, Faden DL, Cullen J, Abadie J, Kwon EM, Gr\u0026ouml;ne A, et al. The MTAP-CDKN2A Locus Confers Susceptibility to a Naturally Occurring Canine Cancer. Cancer Epidemiol Biomarkers Prev. 2012;21(7):1019\u0026ndash;27.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eKaryadi DM, Karlins E, Decker B, vonHoldt BM, Carpintero-Ramirez G, Parker HG, Wayne RK, Ostrander EA. A Copy Number Variant at the KITLG Locus Likely Confers Risk for Canine Squamous Cell Carcinoma of the Digit. PLoS Genet. 2013;9(3):e1003409.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eKarlsson EK, Sigurdsson S, Ivansson E, Thomas R, Elvers I, Wright J, Howald C, Tonomura N, Perloski M, Swofford R, et al. Genome-wide analyses implicate 33 loci in heritable dog osteosarcoma, including regulatory variants near CDKN2A/B. Genome Biol. 2013;14(12):R132.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eSchiffman JD, Breen M. Comparative oncology: what dogs and other species can teach us about humans with cancer. Philosophical Trans Royal Soc B: Biol Sci 2015, 370(1673).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eKoch Z, Graves JL, Annan S, Frankino PA, Nelson M, McMahon JE, Tovar AP, Chen HC, Halioua-Haubold CL, Peloquin M. Multi-omic analysis of canine aging uncovers conserved aging pathways. \u003cem\u003eGeroscience\u003c/em\u003e 2025.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eDas S, Schlemmer SN, Idate R, Lana SE, Regan DP, Thamm DH, Duval DL. Comparative genomic and clinicopathological analysis uncovers contrasting molecular profiles of canine and human thyroid carcinomas. Commun Biol. 2025;9(1):4.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eZhang S-J, Scarsbrook L, Li H, Carmagnini A, Charlton S, Feuerborn T, Boeskorov G, Chen G, Deom J-M, Dimopoulos EA, et al. Genomic evidence for the Holocene codispersal of dogs and humans across Eastern Eurasia. Science. 2025;390(6774):735\u0026ndash;40.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eBoyko AR, Quignon P, Li L, Schoenebeck JJ, Degenhardt JD, Lohmueller KE, Zhao K, Brisbin A, Parker HG, vonHoldt BM, et al. A Simple Genetic Architecture Underlies Morphological Variation in Dogs. PLoS Biol. 2010;8(8):e1000451.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003evonHoldt BM, Pollinger JP, Lohmueller KE, Han E, Parker HG, Quignon P, Degenhardt JD, Boyko AR, Earl DA, Auton A, et al. Genome-wide SNP and haplotype analyses reveal a rich history underlying dog domestication. Nature. 2010;464(7290):898\u0026ndash;902.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eOstrander EA, Wayne RK, Freedman AH, Davis BW. Demographic history, selection and functional diversity of the canine genome. Nat Rev Genet. 2017;18(12):705\u0026ndash;20.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eParker HG, Dreger DL, Rimbault M, Davis BW, Mullen AB, Carpintero-Ramirez G, Ostrander EA. Genomic Analyses Reveal the Influence of Geographic Origin, Migration, and Hybridization on Modern Dog Breed Development. Cell Rep. 2017;19(4):697\u0026ndash;708.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eDutrow EV, Serpell JA, Ostrander EA. Domestic dog lineages reveal genetic drivers of behavioral diversification. Cell. 2022;185(25):4737\u0026ndash;e47554718.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eShannon LM, Boyko RH, Castelhano M, Corey E, Hayward JJ, McLean C, White ME, Abi Said M, Anita BA, Bondjengo NI et al. Genetic structure in village dogs reveals a Central Asian domestication origin. \u003cem\u003eProceedings of the National Academy of Sciences\u003c/em\u003e 2015, 112(44):13639\u0026ndash;13644.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003ePatterson N, Price AL, Reich D. Population Structure and Eigenanalysis. PLoS Genet. 2006;2(12):e190.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMoon KR, van Dijk D, Wang Z, Gigante S, Burkhardt DB, Chen WS, Yim K, Elzen Avd, Hirn MJ, Coifman RR, et al. Visualizing structure and transitions in high-dimensional biological data. Nat Biotechnol. 2019;37(12):1482\u0026ndash;92.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eStreet K, Risso D, Fletcher RB, Das D, Ngai J, Yosef N, Purdom E, Dudoit S. Slingshot: cell lineage and pseudotime inference for single-cell transcriptomics. BMC Genomics. 2018;19(1):477.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eChang CC, Chow CC, Tellier LCAM, Vattikuti S, Purcell SM, Lee JJ. Second-generation PLINK: rising to the challenge of larger and richer datasets. \u003cem\u003eGigaScience\u003c/em\u003e 2015, 4(1):s13742-13015-10047-13748.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eAlexander DH, Novembre J, Lange K. Fast model-based estimation of ancestry in unrelated individuals. Genome Res. 2009;19(9):1655\u0026ndash;64.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eChen S, Lei C, Zhao X, Pan Y, Lu D, Xu S. AncestryPainter 2.0: Visualizing Ancestry Composition and Admixture History Graph. Genome Biol Evol. 2024;16(11):evae249.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eWeir BS, Cockerham CC, ESTIMATING F-STATISTICS FOR THE ANALYSIS OF POPULATION STRUCTURE. Evolution. 1984;38(6):1358\u0026ndash;70.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003ePickrell JK, Pritchard JK. Inference of population splits and mixtures from genome-wide allele frequency data. PLoS Genet. 2012;8(11):e1002967.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eYang Q, Chen J, Nie S, Liu C, Deng H, He G. Fine-scale biogeographical ancestry inference in Southeast and East Asians via high-efficiency markers and machine learning approaches. Front Ecol Evol 2025, Volume 13\u0026ndash;2025.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eWang GD, Cheng LG, Fan RX, Irwin DM, Tang SS, Peng JG, Zhang YP. Signature of balancing selection at the MC1R gene in Kunming dog populations. PLoS ONE. 2013;8(2):e55469.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003ePongr\u0026aacute;cz P, Dobos P. Behavioural differences and similarities between dog breeds: proposing an ecologically valid approach for canine behavioural research. Biol Rev. 2025;100(1):68\u0026ndash;84.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLarson G, Karlsson EK, Perri A, Webster MT, Ho SY, Peters J, Stahl PW, Piper PJ, Lingaas F, Fredholm M, et al. Rethinking dog domestication by integrating genetics, archeology, and biogeography. Proc Natl Acad Sci U S A. 2012;109(23):8878\u0026ndash;83.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eParker HG, Kim LV, Sutter NB, Carlson S, Lorentzen TD, Malek TB, Johnson GS, DeFrance HB, Ostrander EA, Kruglyak L. Genetic Structure of the Purebred Domestic Dog. Science. 2004;304(5674):1160\u0026ndash;4.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLetko A, Minor KM, Jagannathan V, Seefried FR, Mickelson JR, Oliehoek P, Dr\u0026ouml;gem\u0026uuml;ller C. Genomic diversity and population structure of the Leonberger dog breed. Genet Selection Evol. 2020;52(1):61.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eBigi D, Marelli SP, Liotta L, Frattini S, Talenti A, Pagnacco G, Polli M, Crepaldi P. Investigating the population structure and genetic differentiation of livestock guard dog breeds. Animal. 2018;12(10):2009\u0026ndash;16.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eSu\u0026aacute;rez NM, Betancor E, Fregel R, Pestano J. Genetic characterization, at the mitochondrial and nuclear DNA levels, of five Canary Island dog breeds. Anim Genet. 2013;44(4):432\u0026ndash;41.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eOsiak-Wicha C, Kras K, Arciszewski MB. Comparative Analysis of Muscle Fibers in Selected Muscles of Working and Companion Dog Breeds. \u003cem\u003eAnimals (Basel)\u003c/em\u003e 2024, 14(24).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eAcevedo LM, Rivero JL. New insights into skeletal muscle fibre types in the dog with particular focus towards hybrid myosin phenotypes. Cell Tissue Res. 2006;323(2):283\u0026ndash;303.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eTonomura N, Elvers I, Thomas R, Megquier K, Turner-Maier J, Howald C, Sarver AL, Swofford R, Frantz AM, Ito D, et al. Genome-wide association study identifies shared risk loci common to two malignancies in golden retrievers. PLoS Genet. 2015;11(2):e1004922.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eOhmori K, Nishikawa S, Oku K, Oida K, Amagai Y, Kajiwara N, Jung K, Matsuda A, Tanaka A, Matsuda H. Circadian rhythms and the effect of glucocorticoids on expression of the clock gene period1 in canine peripheral blood mononuclear cells. Vet J. 2013;196(3):402\u0026ndash;7.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eWillet CE, Makara M, Reppas G, Tsoukalas G, Malik R, Haase B, Wade CM. Canine disorder mirrors human disease: exonic deletion in HES7 causes autosomal recessive spondylocostal dysostosis in miniature Schnauzer dogs. PLoS ONE. 2015;10(2):e0117055.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eAlves JC, Dos Santos AM, Fernandes \u0026Acirc;D. Evaluation of the effect of mesotherapy in the management of back pain in police working dog. Vet Anaesth Analg. 2018;45(1):123\u0026ndash;8.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eEvans RI, Herbold JR, Bradshaw BS, Moore GE. Causes for discharge of military working dogs from service: 268 cases (2000\u0026ndash;2004). J Am Vet Med Assoc. 2007;231(8):1215\u0026ndash;20.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eWorth AJ, Sandford M, Gibson B, Stratton R, Erceg V, Bridges J, Jones B. Causes of loss or retirement from active duty for New Zealand police German shepherd dogs. Anim Welf. 2013;22(2):167\u0026ndash;74.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLeroy G. Genetic diversity, inbreeding and breeding practices in dogs: results from pedigree analyses. Vet J. 2011;189(2):177\u0026ndash;82.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eKropatsch R, Melis C, Stronen AV, Jensen H, Epplen JT. Molecular Genetics of Sex Identification, Breed Ancestry and Polydactyly in the Norwegian Lundehund Breed. J Hered. 2015;106(4):403\u0026ndash;6.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMellanby RJ, Ogden R, Clements DN, French AT, Gow AG, Powell R, Corcoran B, Schoeman JP, Summers KM. Population structure and genetic heterogeneity in popular dog breeds in the UK. Vet J. 2013;196(1):92\u0026ndash;7.\u003c/span\u003e\u003c/li\u003e\u003c/ol\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":false,"hideJournal":false,"highlight":"","institution":"","isAcceptedByJournal":false,"isAuthorSuppliedPdf":false,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":false,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"bmc-genomics","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"gics","sideBox":"Learn more about [BMC Genomics](http://bmcgenomics.biomedcentral.com/)","snPcode":"","submissionUrl":"https://www.editorialmanager.com/gics","title":"BMC Genomics","twitterHandle":"#BMCGenomics","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"em","reportingPortfolio":"BMC Series","inReviewEnabled":true,"inReviewRevisionsEnabled":true},"keywords":"Domestic dogs, Population structure, Genomic homozygosity, Functional differentiation, Ancestry-informative markers","lastPublishedDoi":"10.21203/rs.3.rs-9303752/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-9303752/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003cp\u003eDomestic dog breeds exhibit extensive phenotypic and functional diversity shaped by long-term artificial selection. However, the genomic processes underlying lineage divergence and functional differentiation between working and companion dogs remain incompletely understood. Here, we analysed over 5,000 genome-wide SNP data to investigate population structure, evolutionary dynamics, and functional genomic differentiation across multiple dog breeds. Genome-wide dimensionality reduction revealed a star-like genetic structure radiating from gray wolves, indicating directional divergence of modern breeds along partially independent trajectories. We further identified nine major lineages broadly corresponding to functional breed groups. Genomic analyses showed that modern purebred dogs exhibit markedly elevated genomic inbreeding compared with village dogs and wolves, and population-level \u003cem\u003eF\u003c/em\u003e\u003csub\u003e\u003cem\u003eROH\u003c/em\u003e\u003c/sub\u003e was strongly positively correlated with mean pairwise \u003cem\u003eF\u003c/em\u003e\u003csub\u003e\u003cem\u003eST\u003c/em\u003e\u003c/sub\u003e among breeds, suggesting that intensive artificial breeding has simultaneously increased within-breed homozygosity and between-breed genetic divergence. Comparative analyses between working and companion dogs revealed significant enrichment of muscle-related pathways, including actin cytoskeleton, myofibril, and contractile muscle fiber, with multiple myosin family genes forming a distinct functional module. Finally, we developed a compact breed identification panel based on ancestry-informative SNPs, in which a 200-SNP panel accurately distinguished eight target breeds. These findings provide insights into the genomic basis of breed diversification and functional specialization in domestic dogs.\u003c/p\u003e","manuscriptTitle":"Genome-Wide Analyses of Breed Differentiation and Functional Specialization in Domestic Dogs","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2026-05-05 00:34:34","doi":"10.21203/rs.3.rs-9303752/v1","editorialEvents":[{"type":"communityComments","content":0},{"type":"editorInvitedReview","content":"","date":"2026-05-13T16:50:47+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"157900925806137995295394787913024933974","date":"2026-04-28T14:40:55+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"133050867268475871750895987508422457997","date":"2026-04-28T12:23:29+00:00","index":"hide","fulltext":""},{"type":"editorInvitedReview","content":"","date":"2026-04-28T10:38:48+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"204607689992188524830101414788150902930","date":"2026-04-25T08:19:11+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"66246038850967845171835251724170446111","date":"2026-04-25T07:55:59+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"116582984816044214669026580276508520070","date":"2026-04-23T11:40:17+00:00","index":"hide","fulltext":""},{"type":"reviewersInvited","content":"","date":"2026-04-23T06:35:07+00:00","index":"","fulltext":""},{"type":"editorAssigned","content":"","date":"2026-04-23T06:31:34+00:00","index":"","fulltext":""},{"type":"editorInvited","content":"","date":"2026-04-13T10:27:42+00:00","index":"","fulltext":""},{"type":"checksComplete","content":"","date":"2026-04-13T07:46:55+00:00","index":"","fulltext":""},{"type":"submitted","content":"BMC Genomics","date":"2026-04-13T01:42:44+00:00","index":"","fulltext":""}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"bmc-genomics","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"gics","sideBox":"Learn more about [BMC Genomics](http://bmcgenomics.biomedcentral.com/)","snPcode":"","submissionUrl":"https://www.editorialmanager.com/gics","title":"BMC Genomics","twitterHandle":"#BMCGenomics","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"em","reportingPortfolio":"BMC Series","inReviewEnabled":true,"inReviewRevisionsEnabled":true}}],"origin":"","ownerIdentity":"83e23491-2205-413d-81e3-cb6c40e382fc","owner":[],"postedDate":"May 5th, 2026","published":true,"recentEditorialEvents":[{"type":"editorInvitedReview","content":"","date":"2026-05-13T16:50:47+00:00","index":70,"fulltext":""}],"rejectedJournal":[],"revision":"","amendment":"","status":"under-review","subjectAreas":[],"tags":[],"updatedAt":"2026-05-05T00:34:34+00:00","versionOfRecord":[],"versionCreatedAt":"2026-05-05 00:34:34","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-9303752","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-9303752","identity":"rs-9303752","version":["v1"]},"buildId":"XKTyCvWXoU3ODBz1xrDgd","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}