Genomewide Prediction of Maize Performance
Cheap and abundant DNA markers have allowed the development of methods to predict the performance of maize lines and hybrids on the basis of DNA fingerprint data. In particular, genomewide prediction entails the development of prediction equations from a training data set of maize lines or hybrids that have been genotyped and evaluated for field performance (i.e., phenotyped). The prediction models are then used to assess the performance of new candidates that have been genotyped but not phenotyped. In 2011, this research group gained access to genotypic and phenotypic data, worth millions of dollars, from a maize breeding company. Such data are allowing them to empirically test different genomewide-prediction models and breeding schemes. Resources at MSI are vital for accomplishing this research goal. Work from 2011 to 2014 has led to a useful procedure to predict maize performance on the basis of historical phenotypic data that are routinely available, along with newly generated marker data. While imputation of marker data for up to 1,000 single nucleotide polymorphism loci is useful, data from next-generation sequencing efforts are not necessary for the prediction of maize performance.
Return to this PI's main page.