The objectives were to assess the efficacy of genomic selection in pure line breeding, using a simulated dataset, the significance of several factors, including model updating, selection intensity, early generation (F2) selection, dominance, and the presence of major-effect genes (QTLs). The simulated genome included 1000 biallelic genes and 49,825 SNPs, distributed on 10 chromosomes of 100 cM. We used genomic selection with partial phenotyping over generations and other scenarios. The efficacy of genomic selection was based on total realized genetic gain and probability of selecting superior pure lines. The results showed that genomic selection with model updating maximized the probability of selecting superior F8 progeny and provided the highest total genetic gain, comparable to selection based on the true genotypic value. Larger training sets (achieved through model updating) and higher selection intensity were key factors affecting the development of elite pure lines. Dominance did not significantly affect genomic selection efficiency. Including QTLs increased genomic selection efficiency. Direct selection imposed within the F2 generation was no more effective than selection started in F3. All selection methods provided a high decrease in the genotypic variance at F8. The realized genetic gains per cycle were positively correlated with the prediction accuracies.
Silva et al. (Tue,) studied this question.