Genomic Selection By Pooled Genotyping

Genomic selection (GS) has been successfully applied in the dairy cattle industry and its theoretical benefits, especially for species with long breeding cycles, is unquestionable. Yet, the wide incorporation of GS has been slow in most crop species, and has largely remained restricted to an experimental stage. A significant limitation to the extensive implementation of GS remains the high cost and logistic complexity of genotyping large populations. Additionally, breeders are understandably reluctant to rely exclusively on selections performed by GS without additional testing. As a consequence, GS will likely be used initially as a screening tool, to improve the probability of identifying the best genotypes. Next, selected lines will be tested in the field to obtain additional data and to confirm their superior properties. Here we present an innovative approach to apply GS in breeding program that fits this scheme, significantly decreasing the cost of genotyping while simplifying the genetic data acquisition process. The approach relies on sequencing-based genotyping of family pools of DNA. For each pooled sample, family-specific allele frequencies are estimated and combined with family trait means to create prediction models. Our results using real data obtained from a breeding population revealed that accuracies obtained from pooled and individual data are equivalent. To verify the consistency of these results under different population and trait scenarios, we expanded this analysis to simulated data, which confirmed the value of this method in breeding programs of outcrossing species.