عنوان

بکارگیری مدل های آماری انتخاب ژنومی و آنالیز ‮‭QTL‬ در برنامه های اصلاح نژادی گوسفند

Text of Note

The aim of this study was to determine the effect of marker density, level of heritability, number of QTL and effective population size (Ne) under two different QTL effect distributions on genomic accuracy with sheep simulated and real data. Thereby, a genome with three chromosomes, 100 cM each simulated with QMsim software. The real data prepared from International Sheep Genomics Consortium (ISGC) website. For simulated data, a trait was simulated with three heritability lavels (0.10, 0.30 and 0.50), three different marker densities (1000, 2000 and 3000 markers) and numbers of QTL were assumed to be either 100, 200 or 300 in three Ne stages (50, 100 and 200) with two different distributions for the QTL effect which were uniform and gamma (= 1.66, =0.4). The true and estimated genomic breeding value accuracy which obtained by of GBLUP, BayesA, BayesB and Bayes LASSO were compared. The results showe that the estimated genomic accuracy by all four statistical methods in all traits is a very high correlation with the true genomic accuracy (r > 0.7). The BayesB produced estimates with higher accuracies by reducing the number of QTL, increasing heritability, increasing the number of SNP and the gamma QTL effects distribution than the other three studing methods. On the other hand, in trait with high QTL numbers, superiority of Bayesian estimates accuracy were not visible and even in some cases GBLUP method more accurate estimation showed. Also passed from generation, the correlation between true and the genomic estimated breeding value in target generations decreased. Adaptation of simulation results with real data results showed that increasing the number of markers and heritability with reducing the number of QTL affecting the trait, can be effective for enhancing the accuracy of genomic breeding values