Statistical modeling for selecting a superior working dog using microsatellite loci

Abstract

Dogs (Canis familiaris) have been around humans for 12,000 years or more. Genetic background or external environment allow many dog breeds to be trained to perform highly specialized tasks such as detecting drugs and guiding visually impaired people. In this study, we selected 50 Sapsaree dogs, a Korean breed that has the potential as a working dog. They underwent training and were scored during the in-training examination; given their scores, they were divided into pass or fail groups. We analyzed genetic difference between pass and fail individuals using genotyping of 13 microsatellite loci. The mean number of alleles, allelic richness, and observed heterozygosity of the pass group were 4.077, 4.061, and 0.489, respectively, whereas those of the fail group were 4.154, 4.138, and 0.465, respectively. Using the genetic information, we constructed statistical modeling. The results of both logistic regression analysis and decision tree indicated that the superiority of Sapsaree was determined by 166 repeats of bases (166 allele) and 164 repeats (164 allele) at TAT locus and 198 repeats (198 allele) of DRPLA gene. The results presented herein indicate that these allelic differences between the pass and fail group can be a good biomarker for selection of superior Sapsaree individuals, and our statistical modeling could provide the standard for selection of working dogs.