Factors affecting the accuracy of genomic selection for quality traits within a biparental wheat population (CROSBI ID 724731)
Prilog sa skupa u zborniku | sažetak izlaganja sa skupa | međunarodna recenzija
Podaci o odgovornosti
Plavšin, Ivana ; Gunjača, Jerko ; Novoselović, Dario
engleski
Factors affecting the accuracy of genomic selection for quality traits within a biparental wheat population
Genomic selection is one of the recently deployed approaches to breeding which aims to reduce the potential costs of phenotyping that occur in the classical breeding process. The basic principle of genomic selection consists of the estimation of marker effects using phenotypic and genotypic data of a training population that will be later used to calculate genomic-estimated breeding values (GEBV) of a validation population. The success of genomic selection depends on the obtained prediction accuracy which can be affected by many factors. The present research aimed to assess the impact of marker density, training population size, and chosen model on prediction accuracy when predicting quality traits within a biparental population. The population comprised of 153 recombinant inbred lines (RILs) derived from a cross between Monika and Golubica cultivars. The experiment was conducted for 3 consecutive years at Osijek and Slavonski Brod (Croatia) and predictions were done across all six environments. Obtained phenotypic data included grain protein content (GPC), test weight (TW), and time required for optimal dough development (midline peak time ; MPT). To assess the effect of marker density on prediction accuracy two sizes of marker dataset (NM = 2231 and NM = 1123) were used in predictions, while the assessment of training population (TP) size was done using 50%, 60%, 70% and 80% of the population as TP (70, 83, 97, 111 RILs). Prediction models used in the present study were RR-BLUP, BayesLASSO, Elastic Net, and Random Forest. Broad-sense heritabilities for all traits were high (H2 ≥ 0.78). A positive impact of the increase in population size and marker density was noticed for all investigated traits. The highest mean prediction accuracies were obtained using BayesLASSO, RR-BLUP, and Random Forest for GPC, TW, and MPT, respectively. In overall, the highest prediction accuracies were observed for MPT (r = 0.43 – 0.57), followed by GPC (r = 0.39 – 0.49) while genomic selection for TW seems to be least effective (r = 0.29 – 0.37).
wheat ; quality ; biparental population ; genomic selection ; prediction accuracy
nije evidentirano
nije evidentirano
nije evidentirano
nije evidentirano
nije evidentirano
nije evidentirano
Podaci o prilogu
33-33.
2022.
objavljeno
Podaci o matičnoj publikaciji
Book of Abstracts of International Conference on Biodiversity and Molecular Plant Breeding
Podaci o skupu
International Conference on Biodiversity and Molecular Plant Breeding
predavanje
02.10.2022-06.10.2022
Novigrad, Hrvatska, 02.10.2022. - 06.10.2022