The importance of disease incidence rate on accuracy of genomic selection methhods

In order to o predict genomiic accuracy of binary traits based on different rates of disease incidence, two machine learning algorithms including Boosting and Random Forest (RF) as well as threshold BayesA (TBBA) as well as genomic genomic best linear unbiased prediction (GBLUP) were applied. For this purpose, the predictive performance of each of these models was evaluated in different co mbinationsof genomic architecture using imputed (95, 75 and 50% missing rate) and their original 50k genotypes.The highest predictive ability belonged to GBLUP, RF, Boosting and TBA methods were when the rate of disease incidence into the training set was 16 percent.In general, for different genomic architectures, the Boosting method performed better than the TBA, GBLUP and RF methods for the accuracy of genomic prediction under all scenarios and proportions of the markers set being imputed