The emergence of the family Equidae has a longer evolutionary history than the emergence of Homo sapiens. The evolution of horses occurred over ~50 million years. At first, people considered horses to be hunting targets, but those who found out their abilities began to domesticate them ^[1][2][3][1,2,3]. The horse was domesticated and used for farming, transportation, food, and war purposes. After the Middle Ages, domestic horses were widely distributed and were commonly used for human hobbies such as riding and racing. As a result, the Thoroughbred horse has become widespread as the most representative horse today through pedigree management. The Thoroughbred horse has continued to manage superior objects in the same way as single nucleotide polymorphism (SNP) and restriction fragment length polymorphism (RFLP) [4]. Recently, several types of research using next-generation sequencing (NGS) technology, such as whole-genome sequencing, target sequencing, and RNA sequencing, to characterize genetic features of different horses have been conducted ^{[5][6][7][8][9]}[5,6,7,8,9]. However, genetic research on Jeju horses using NGS has been relatively less studied.

Jeju horse is a general term for the ponies that grow wild on Jeju Island. It was designated as National Monument number 347 on 8 February 1986, in Korea. It is estimated to be a hybrid of the pony that used to live on Jeju Island and the Mongolian horse that flowed into the island from the late 1200s to the 1300s ^[7][10][7,10]. The average body height of an adult Jeju horse over five years old is 117 cm for females and 115.3 cm for males, which is smaller than a mixed horse or modern horse (crossbreed or improved breed), and differs genetically from other horses [11]. The fur of the Jeju horse is brown, reddish-brown, gray, black, light yellow, and stained, with the brown individuals being the most common, followed by reddish-brown. In addition, the Jeju horse has a low front, high backside, medium body, and a large chest ratio, which is a typical body shape suitable for a sumpter (Figure 1).

Jeju horses have been close to extinction in the past. However, the protection afforded by the Endangered Species Act means that the population of Jeju horses has rebounded today, with more than 5000 Jeju horses breeding [12]. Most studies on Jeju horses are morphological compared to Thoroughbred horses. Furthermore, the most basic genetic studies for the academic establishment of Jeju horses are far from insufficient [13].

In this entry, we performed whole-genome resequencing (WGRS) of five domesticated Jeju horses and one Thoroughbred. Using the horse reference genome data, we determined a significant number of SNPs and insertion/deletion (INDEL) throughout the genome. All the structure variants were annotated, particularly resulting in nonsynonymous mutations that can be further used as genetic markers that would predict phenotypic variation in patterns of interest in Jeju horses. Interestingly, the results of the WGRS comparison indicated that the eqCD1a6 gene contains signatures of positive natural selection in Jeju horses and is thought to be the result of environmental adaptation. In addition, we confirmed the Gene Ontology (GO) term of Jeju horses through GO analysis and found that there are many correlations in genes related to the heart and nerves. These results are indirect evidence that Jeju horses show physical differences from Thoroughbred horses.

2.1. Sequencing, Read Mapping, and Genomic Variant Detection

To detect Jeju horse-specific genomic variation, we compared the genomes of Jeju horses, which have been geographically isolated for a long time, with the horse reference genome. In order to obtain more accurate results, we generated additional genome data of a Thoroughbred horse breed in the same species as the horse reference genome. We performed WGRS on five Jeju horses and one Thoroughbred horse using the Illumina HiSeq 2500 platform and obtained a total of 696,089,910 raw reads. The raw reads were trimmed and deduplicated via Sickle and Picard, resulting in an average of 609,556,376 clean reads. Next, the clean reads were mapped to the horse reference genome using BWA, and finally, the average read depth of the six horses ranged from 31.48x to 46.01x (an average of ~35.79x, Table 1). ^a Total reads: The total number of reads generated. ^b The number of reads after trimming and deduplication with Sickle and Picard; (%) = No. of Clean reads/No. of Total reads. ^c The number of reads mapped to the reference using BWA mapping tool; (%) = No. of Mapped reads/No. of Clean reads. To identify Jeju horse-specific structural variation (SV), we compared the genome data obtained from five Jeju horses with the horse reference genome using the GATK tool and the variant annotations in VCF files were created with SnpEff. As a result, we found an average of 6,686,678 SNPs, 436,460 insertions, and 456,249 deletions in the Jeju horse (Table 2). 18]. The previously registered SNPs consisted of 21,443,129 dbSNPs, 5,008,750 Ensembl, and 1,163,466 Broad Institute. Finally, a total of 408,601 Jeju horse-specific SNPs that do not overlap with open access databases were identified (Figure 4). The 408,601 Jeju horse-specific SNPs were divided into 94,192 homozygous and 314,409 heterozygous. For the 113,498 Jeju horse-specific insertions, they were divided into 85,394 homozygous and 28,104 heterozygous. For the 114,751 Jeju horse-specific deletions, they were divided into 75,115 homozygous and 39,636 heterozygous. Among the identified SVs, we analyzed the loci of those corresponding to homozygous SVs (94,192 homozygous SNPs, 85,394 homozygous insertions, 75,115 homozygous deletions) in all five Jeju horses. This showed that most SVs resided in intergenic regions (an average of 64%), followed by an intron, upstream region, and exon (Table 3). ^a Homo = homozygous; Hetero = heterozygous; INS = insertion; DEL = deletion. Next, we analyzed the SV positions shared by five Jeju horses to find Jeju horse-specific SVs. As a result, 2,758,242 SNPs, 233,819 insertions, and 240,738 deletions were identified as SVs shared by five Jeju horses (Figure 2). The SV results shared by five Jeju horses were compared with the WGRS data of one Thoroughbred horse to further reduce the number of Jeju horse-specific SVs by obtaining 1,244,064 SNPs, 113,498 insertions, and 114,751 deletions (Figure 3). Furthermore, Jeju horse-specific SNPs were compared with SNP data from open-access databases (dbSNP, Ensembl, and Broad Institute) ^[14][

2.2. Functional Annotation of Nonsynonymous

Among the numerous Jeju horse-specific SVs, we focused on nonsynonymous SNPs (724) and INDELs (564 insertions and 879 deletions) present in genic regions that may have a more functional impact. We performed GO term analysis using the ClueGO plugin of Cytoscape software ^[15][19] on 788 genes containing 2167 SVs (nonsynonymous SNPs and INDELs). As a result, 106 out of 788 genes were correlated with 13 GO categories . The 13 GO categories were mainly involved in cardiac development/blood circulation and neurodevelopment/hormone secretion. The reason why these two groups are characterized is probably due to the external differences between the Jeju horse and Thoroughbred horse. Thoroughbred horses have been gradually improved to be faster by humans. As a result, they have a body suitable for high speed, but their endurance was weakened. In contrast, Jeju horses are well known for their endurance. Due to these differences, we hypothesize that Jeju horses and Thoroughbred horses show significant differences in cardiac development/blood circulation and neurodevelopment/hormone secretion (Figure 5). Interestingly, we found that 275 SNPs and 21 INDELs common in five Jeju horses exist in the eqCD1a6 gene (Table 5). The function of the CD1 gene family is unknown, but recently it has been known to be involved in immunity to Rhodococcus equi. R. equi is associated with Mycobacterium tuberculosis and is structurally similar to the nocardioform actinomycete bacterium ^[16][20]. M. tuberculosis is known as a cause of pulmonary tuberculosis in humans. In contrast, in young horses, R. equi is known to be a life-threatening pathogen that causes pyogranulomatous pneumonia ^[17][21]. Previous studies have reported that most mammals have more than one isoform of the CD1 gene ^[18][22]. In a previous study, 13 complete eqCD1 genes were identified in the horse genome, and they were largely divided into eqCD1a, eqCD1b, eqCD1c, eqCD1d, and eqCD1e. Among them, eqCD1a is the largest isoform group (eqCD1a1~eqCD1a7) ^[19][23]. The eqCD1a6 gene is 2281 bp in length and consists of six exons that encode the signal peptide, α-1, α-2, α-3, the transmembrane region, and the cytoplasmic tail. Of the 275 SNPs identified in Jeju horses, 51 SNPs were located in the exon regions of eqCD1a6 and the most SNPs were found at Exon 2 and Exon 3 (18 and 25 SNPs, respectively) (Figure 6). We compared the amino acid sequence of the eqCD1a6 of Jeju horses with its counterpart in the horse reference genome. As a result, the eqCD1a6 gene has accumulated 37 nonsynonymous changes, but no stop codons have been found. Most nonsynonymous changes occurred in α-1, α-2, and α-3 (14, 16, and 5 nonsynonymous substitutions, respectively) . In order to confirm the correct Jeju horse-specific SNP of the eqCD1a6 gene, we used additional DNA samples of 35 horses (Jeju horse 15, Halla horse 3, and Thoroughbred (raised in Jeju Island) 17), which were used to PCR the exon region of eqCD1a6 and confirmed by Sanger sequencing. As a result, out of 51 SNPs, we identified 36 SNPs unique to Jeju horses, which most Jeju horse samples have in common in the Jeju horse genome. We conducted a dN/dS ratio analysis based on Jeju horse-specific SNP data ^[20][24] (Figure 7). The dN/dS ratios estimate the evolution rate by considering synonymous and nonsynonymous variations. The eqCD1a6 gene in Jeju horses appears to have been a positive selection to escape from the current stage when compared to Thoroughbred horses.

2.4. Genotyping Assay for Molecular Markers

Korea is conducting pedigree management to protect species and manage individual Jeju horses, and species identification is performed through the appearance of Jeju horses and various genetic tests. However, sometimes inaccurate results are obtained because Jeju horse genome data are not compared with other Jeju horses. We proceeded with the development of a molecular marker. Based on the Jeju horse-specific CD1a6 obtained through the research results, we expected that the development of a species-specific molecular marker would enable a more accurate and faster test, and digital PCR was applied to this. Digital PCR has become accessible and easier to handle due to recent advances in technology and the diversification of equipment. The LOAA equipment used in this study is a semiconductor method that is different from the existing droplet digital PCR and has ultra-fast, ultra-light, ultra-compact, and ultra-sensitive features. We designed the probe and primer set based on the SNPs of the eqCD1a6 gene with the most variation between the Jeju horse and Thoroughbred. For the sample, 10 gDNA samples were used for each Jeju horse and Thoroughbred. As a result, it was shown that the unique aspect of the Jeju horse was confirmed through the molecular marker, and the accuracy reached 80% (Figure 8) In this study, since only one probe was used, it showed an accuracy of 80%. However, we predict that designing Jeju horse-specific probes based on sequence information of various SNPs will be a more accurate species molecular marker. Based on these results, it is thought that this experimental method can be applied to various fields. Furthermore, if digital PCR is as easy to operate and lightweight as LOAA, it is expected that digital PCR is expected to be fully utilized in point of care testing (PoCT).