نوع مقاله : مقاله پژوهشی
نویسندگان
1 محقق مستقل (دکتری ژنتیک و اصلاح نژاد دام)، دانشکده کشاورزی، گروه علوم دامی، دانشگاه کردستان، کردستان، ایران
2 گروه علوم دامی، واحد آستارا، دانشگاه آزاد اسلامی، آستارا، ایران.
چکیده
کلیدواژه‌ها
موضوعات
عنوان مقاله [English]
نویسندگان [English]
The purpose of this study was to investigate the effect of different rates marker genotyping error and the type of mating and selection design (breeding value and phenotypic) on the accuracy of genomic prediction assessment under different levels of heritability (0.05, 0.1 and 0.3) and marker density (500, 1000 and 1500) by simulation in threshold trait. The genome consisted of two chromosomes, each 100 cM, and 125 QTLs were randomly distributed on each chromosome. In order to simulation a threshold trait, 20 percent of the top-level phenotypes were considered to be 2, and the rest were considered as 1. Genomic breeding value was predicted using marker effects estimated by Bayes B statistical method. Comparison of the accuracy of genomic evaluations showed that selection and mating designs of breeding value was more accurate than the selection and mating designs of phenotypic. The accuracy of genomic prediction decreased with increasing marker genotyping error in both selection and mating designs of breeding value and phenotypic. The results showed that with increasing the percentage of marker genotyping error, increasing the number of markers leads to increasing the accuracy of genomic breeding value prediction.
کلیدواژه‌ها [English]
Extended Abstract
Introduction
Many traits recorded in domestic species, including the litter size of large mammals, the extent of calving difficulty, and resistance to diseases, exhibit a discrete distribution of phenotypes, commonly referred to as threshold traits. Genetic progress in such traits depends on genetic diversity in the population, selection intensity, accuracy of prediction and generation interval. Genomic selection is a novel method to improve quantitative traits in plant and animal breeding. The factors affecting the accuracy of genomic prediction are the heritability of the interested trait, number of individuals in the reference population, the extent of relationships between selection candidates and the reference population, the accuracy of estimated marker effects, linkage disequilibrium between markers and quantitative trait loci (QTL), the distribution of QTL effects and marker genotyping error. The reported range of marker genotyping error falls between 0.1% and 15%, which may significantly influence genomic prediction accuracy. Hence, the objective of this study is to explore the impact of varying heritability levels, marker intensities, rates of marker genotyping error, as well as different types of selection and mating strategies, on the precision of genomic prediction for a threshold trait
Materials and methods
To conduct a comparative analysis of different scenarios, we employed the QMsim software for simulating various datasets. Simulation started with a base population of 1000 animals, including 500 males and 500 females, which randomly mated for subsequent 1000 generations. Subsequently, a random selection of 20 females and 200 males was made from the last historical generation to expand the population size for an additional 10 generations. The training sets comprised individuals from the 8th to the 9th generations, while the validation set consisted of all individuals from the 10th generation. The simulated genome consisted of two chromosomes, each with an equal length of 1 Morgan. Scenarios were established to examine the impact of various factors on the accuracy of genomic prediction. These factors included different rates of marker genotyping error (0%, 4%, 8%, and 12%), different types of mating and selection designs (based on breeding value or phenotype), varying levels of heritability (0.05, 0.1, and 0.3), and different marker densities (500, 1000, and 1500). In this study, a threshold model was employed, and the marker effects were estimated using the Bayesian B methodology. To assess the accuracy of prediction, the correlation between the true and estimated genomic breeding values was calculated.
Results and discussion
The accuracy of predictions, using breeding value as the mating and selection design, ranged from 0.54 to 0.56 for h2 = 0.05, 0.61 to 0.66 for h2 = 0.10, and 0.79 to 0.82 for h2 = 0.30. On the other hand, the prediction accuracies, employing phenotypic as the mating and selection design, varied from 0.63 to 0.69 for h2 = 0.05, 0.69 to 0.72 for h2 = 0.10, and 0.79 to 0.84 for h2 = 0.30. The comparison of these results revealed that the selection and mating designs based on breeding value exhibited higher accuracy compared to the selection and mating designs based on phenotypic traits. The results demonstrated that increasing the number of SNPs from 500 to 1500 led to improved prediction accuracy for both breeding value and phenotypic-based mating and selection designs. However, it was observed that the accuracy of genomic prediction declined as the marker genotyping error increased from zero to 12 percent in all scenarios.
Conclusion
In summary, the findings indicate that increasing the number of markers from 500 to 1500 mitigates the impact of marker genotyping error rates from zero to 12%. Consequently, increasing the number of markers enhances the accuracy of genomic evaluation.
)
