Polyploidization (that is, whole-genome multiplication) among plants is a pivotal contributor to evolution. US researchers have now developed a statistical learning model to predict whether a duplicate gene is likely to be retained after a polyploidization event on the basis of features identified using comparative genomic analyses of wild radish and three other Brassicaceae species. Retained radish duplicates showed substantial divergence in sequence and expression. They also differed markedly from genes that reverted to singleton states; for example, they have a lower rate of evolution, higher network connectivity, and broader and higher expression.