Pedigree-based genome-wide imputation using a low-density amplicon panel for the highly polymorphic Pacific oyster Crassostrea (Magallana) gigas

Abstract

High-density genomic data are instrumental for selective breeding, but the costs associated with these approaches can hinder progress, as is the case for most aquaculture species. A strategy to reduce genotyping costs is to genotype a few select individuals at high-density (e.g., parents, grandparents), and many others at low density (e.g., offspring), then impute genotypes. This has been demonstrated in silico for Pacific oyster Crassostrea (Magallana) gigas but was particularly challenging relative to other species and has never been empirically tested. Here, four families of Pacific oysters, bred via marker-assisted selection for variation at a locus for field survival in an ostreid herpesvirus 1 (OsHV-1)-positive estuary, were exposed to OsHV-1 then genotyped using a low-density amplicon panel (n = 240 individuals). Parents were genotyped with the amplicon panel and by whole-genome resequencing. Offspring genotypes were imputed, and accuracy was determined by comparing against held-out whole-genome data for offspring. Imputation resulted in reduced minor allele frequencies and enriched homozygosity relative to empirical data. An in silico three-generation analysis was used to investigate the effect of deepening the pedigree, resulting in superior concordance in genotypes (GC = 84.5%) and allelic dosage (r = 0.73) compared to two-generation imputation (GC = 75.3%; r = 0.63). Genome-wide associations to OsHV-1 survivorship with imputed data identified significantly associated regions on the expected chromosome 8, but not at the expected position based on previous work, pointing to a potentially more complex genetic architecture for the trait. Our results empirically demonstrate the utility of amplicon panel-based genome-wide imputation in shellfish, and thus enable low-cost selective breeding techniques.