Microhaplotypes generate higher breeding value accuracy compared to SNPs for imputation-based breeding strategies

Abstract

The cost of genotyping remains a persistent barrier to the adoption of genomic selection for many breeding programs. This is particularly prominent in the aquaculture industry because the high number and diversity of cultured species impedes the reduction of genotyping costs through a shared genetic panel (e.g., a high-density SNP array). One solution is to reserve high-density genotyping for key individuals and apply a low-density SNP panel along with pedigree-informed imputation to the remaining individuals. We examined the possibility of further lowering the cost of this strategy by targeting microhaplotypes instead of SNPs in the low-density panel, which could allow smaller panels to be used. We simulated Pacific oyster, eastern oyster, and Atlantic salmon breeding programs for three generations and compared imputation and GEBV accuracy using low-density panels targeting SNPs or microhaplotypes. Panels targeting microhaplotypes yielded higher imputation and GEBV accuracy than that of equally sized panels targeting SNPs. In the Pacific and eastern oyster simulations, close to the maximum imputation and GEBV accuracy was reached when the low-density panel contained 150–250 microhaplotypes or 350–450 SNPs. In the Atlantic salmon simulations, this level of accuracy was reached with low-density panels of 350–450 microhaplotypes or 650–750 SNPs. Using low-density panels targeting microhaplotypes instead of SNPs can reduce the cost of genotyping and thereby make genomic selection feasible for a wider range of programs.