Synbreed: Genomic prediction of hybrid performance

Application project A3

Predicting the performance of untested crosses is of great importance for the efficiency of hybrid breeding programs. Until now, performance prediction was solely based on phenotypic information from relatives, and was therefore very resource intensive with limited accuracy at the same time. A shift of paradigm in plant and animal breeding appears to be very promising based on previous experimental results when using "genomic selection" in breeding of pure-bred lines, suggesting to complement the phenotypic performance prediction by genomic performance prediction or even to replace it.

In the case of crosses, not only additive gene effects, but also dominance, epistasis and epigenetic phenomena have to be taken into account. Therefore, in Synbreed A3, quantitative-genetic models are derived to account for such diverse effects. Appropriate data will be generated for maize, chicken and beef by SNP genotyping and phenotyping following standard practice, and analyzed using statistical methods developed in other Synbreed projects. Accuracy of the genomic performance prediction and the sensitivity of SNP estimates to environmental iufluences are examined using cross-validation methods developed. In addition, simulated data are used to investigate the influence of important factors such as marker density and kinship structure on the quality of the genomic performance prediction.

Based on the results, strategies for optimal integration of genomic performance prediction in breeding programs of the studied species; but also for plant and animal breeding in general will be derived.