Genomic selection utilizes high-density genotype information for both phenotyped individuals and selection candidates to improve genetic gains in a selective breeding program. To reduce genotyping costs, key individuals can be genotyped with a high-density panel while other individuals
are genotyped with a low-density panel. G enotypes for the loci absent from the low-density panel can then be inferred through imputation so that the genomic estimated breeding values (GEBVs) can be calculated using high-density genotypes from all individuals. When designing such a program, one must know the number of loci required in the low-density panel to achieve sufficient imputation accuracy for GEBV calculations.
The number of loci in a low-density panel necessary for imputation depends on several factors including allele frequencies , distribution of loci throughout the genome, size of the genome, and the extent of linkage disequilibrium in the target population . Several of these factors are species specific. We therefore designed a series of simulations to investigate the number of loci that would be required for eastern oysters Crassostrea virginica and Pacific oysters C. gigas .
Our simulated breeding program consisted of 100 crosses per generation with 50 offspring per cross. All broodstock were genotyped with a high- density panel and all offspring were genotyped with a low-density panel. We simulated three generations and recorded both imputation accuracy and GEBV accuracy in each generation.
In the first simulation , founder genomes were generated
based on demographic histories representative of wild oyster populations . Three additional simulations were run with high-density genotypes from either wild oysters or a Pacific oyster breeding program being used to define founder genotypes.
These simulations demonstrated that imputation accuracy with a low-density panel of 250-500 loci was sufficient for GEBV accuracy to be comparable to that obtained with high-density genotypes for all individuals when genotypes for three generations (grandparents, parents, offspring) were available. This occurred beginning in the second generation of selection candidates.
Panels of this size are achievable using currently available
amplicon sequencing methodologies, making this approach immediately available for cost-effective, genomic selection in oysters.