The accuracy of different strategies for building training sets for genomic predictions in segregating soybean populations

• DOI: 10.1002/csc2.20267
• Publication Date: 2020-07-17T13:25:05Z
• Authors: Leandro de Freita...; Roberto Fritsche‐...
• Keywords: Genomic Selection; Trait; Training set; Breeding program; Plant Breeding

Abstract The design of the training set is a key factor in success genomic selection approach. nature line inclusion soybean [ Sorghum bicolor (L.) Moench.] breeding programs highly dynamic, so generating that endures across years and regions challenging. Therefore, we aimed to define best strategies for building sets apply segregating populations traits with different genetic architectures. We used two datasets grain yield (GY) maturity group (MG) from Brazil. Five schemes were tested. In addition, included formed by an optimization algorithm based on predicted error variance. predictions achieved good values both traits, reaching 0.5 some scenarios. scenario changed according trait. Although performance was use full‐sibs MG dataset, GY, advanced lines equivalent. For no improvement predictive ability resulted optimization. Furthermore, same program recommended as continually renewed closely related populations, additional phenotyping needed. On other hand, improve prediction accuracies MG, it necessary less variability but more segregation resolution.