Exploring the Efficacy of Genomic Selection in Rice Using Elite Lines of the International Rice Research Institute's Irrigated Rice Breeding Program

Genomic selection (GS) refers to the process of using genomewide marker data to calculate genome estimated breeding values (GEBVs) among progeny of a breeding population in order to identify plants to advance within a breeding cycle and make rapid gain from selection. Unlike in traditional marker assisted selection, selections are not made based on a subset of "informative" markers, but rather on the results of a model built using a training set of lines that has been both genotyped and phenotyped in order to predict breeding values in related populations using all available marker data, without phenotyping. In crop plants, successful case-studies in maize and small grains have suggested that GS has the potential to increase gain from selection per unit time while decreasing cost by dispensing with some or most phenotyping currently essential to breeding programs.  As yet, however, no theoretical or applied genomic selection research has been published on rice. Here we investigate applying genomic selection to rice breeding using empirical yield data on a subset of elite breeding lines selected from the International Rice Research Institute's irrigated rice breeding program. 366 entries were genotyped using Genotyping-By-Sequencing at Cornell University. Cross-validation experiments will be performed using four years of yield data with 2 seasons collected per year in Los Baños, Philippines, as well as 2 years of yield data collected at an additional 6 locations in Southeast Asia. Five-fold cross validation scenarios are planned to explore the extent to which genomic selection can be successfully used in rice.