Testing frequency-based methods of genic selection in addition to genomic selection in goats

• Genic selection should be dimensioned in reference to genomic selection. • Genetic gain and coancestry rate must be considered simultaneously as genic selection. • Considering allelic frequencies instead of economic weights is a good alternative. • Increasing frequencies of any favourable allele is tested and shown to be possible. • Tested in goat breeding, the approach is extensible to several other species. Genic selection manages specific genes or genomic regions as part of a breeding scheme. Its uncontrolled application can substantially impact genetic gain for an overall objective or inbreeding or both. A traditional approach would consider economic impacts of genic selection. However, in some cases, the economic values attached to genotypes involved in genic selection cannot be determined accurately. This study proposes a method to apply genic selection without assigning economic value to individual genes, that does not have excessive impacts on genetic gain and/or inbreeding of an existing initial breeding scheme. The final breeding scheme is implemented under two major constraints, to be defined by the breeding organisation: maximum allowed change in the annual rate of genetic gain and in the inbreeding rate. Once these constraints are specified, genic selection is optimised within the defined limits. In this context, five methods are considered for increasing the frequencies of the most favourable alleles through genic selection. These methods are tested on a dataset from the French Saanen goat breeding programme producing young artificial insemination bucks, where the genic selection considers six genes with uncertain economic values and highly diverse characteristics. The base situation was to maintain the rate of genetic gain while allowing the inbreeding rate to increase moderately. Simulated annealing was used to obtain the contributions of selected sires, in order to optimise the frequencies of the favourable alleles in the progeny. In comparison to the allele frequencies obtained after running the genomic breeding scheme without genic selection, genic responses were substantial for three genes. Otherwise, genic responses were null or weak (negative responses were forbidden): for one gene due to its initial very high frequency of the favourable allele, and for the other two genes due to their unfavourable associations with other genes. These results suggested that in case of uncertain economic assessment, these methods could be useful in practical breeding schemes.