Untersuchungen zur Genauigkeit einer genomgestützten Zuchtwertschätzung

Abstract

Modern developments in animal breeding in combination with advanced understanding of molecular genetics proceeded in the end of the 20th century. Most genetic gain was achieved in production traits with high/moderate heritabilities. All known selection strategies are limited for traits with low heritabilities or carcass traits. Therefore new developments as genomic selection, using direct genetic information, are of great interest in animal breeding. Goal of this study was to estimate genomic breeding values for individuals, based on the framework of a pig population. Methods used were a BLUP and a resampling approach. Different parameters were tested, different population sizes, heritabilities, selection strategies and haplotypes. The accuracy of genomic breeding values and BLUP breeding values were compared. A population of 300 animals was simulated. This population size has been kept constant for the first 15 generations. In generation 15 the number of animals was increased because of the higher number of offspring (10) per parents. This number (1500) was kept constant for the following generations. A subsample of 300 animals was drawn randomly from generation 16. These animals were used to estimate first marker/haplotype effects. The simulation for all animals based on one chromosome with a length of one morgan containing 100 SNP s, which were randomly spread over the whole sample. A total of 12 SNP s were randomly chosen to become QTL s, whose alleles defined the true breeding value of each animal. Additionally a phenotypic value for each animal was simulated. Based on the data of these 300 animals the haplotype/marker effects were estimated to obtain genomic breeding values for each animal. Results were: