Genomic Predicted cross performance: a tool for optimizing parental combinations in breeding programs

Genomic prediction is an effective method for shortening breeding cycles and accelerating genetic gains. Traditionally, genomic prediction has focused on estimating 'additive' breeding values for individual genotypes. However, for many breeding programmes, predicting the cross-performance of parental combinations may provide greater value.

We present the genomic predicted cross-performance (GPCP) tool, which utilizes a mixed linear model based on additive and directional dominance. This tool is available within the BreedBase environment and as an R package. We assessed its effectiveness against classical genomic estimated breeding values (GEBVs) using simulated traits that exhibit varying dominance effects and on four yam traits. The GPCP tool proved superior to traditional methods for traits with significant dominance effects, effectively identifying optimal parental combinations and enhancing crossing strategies. This article outlines how the tool is implemented and emphasizes situations where predicting cross-performance is more advantageous than depending solely on GEBVs.

The GPCP tool provides a robust solution for predicting cross-performance, offering significant advantages for breeding programmes targeting traits influenced by dominance. It is particularly useful for clonally propagated crops, where inbreeding depression and heterosis are prevalent and reciprocal recurrent selection is impractical.