Assessing accuracy of imputation using different SNP densities as strategy for breeding programs of the fish pacu Piaractus mesopotamicus

Pacu (Piaractus mesopotamicus) is one of the main native fish species for aquaculture in South America. Genomic selection is essential for developing desirable traits in pacu production, but genotyping at high densities of SNP markers can be costly for producers. Cost-effective genotyping strategies involving genotype imputation can facilitate the broader adoption of genomic selection for this species. Then, the main objective of this study was to evaluate imputation accuracies by adopting different SNP densities (9 K, 7 K, 5 K, 2 K, 1 K, and 0.5 K) from the commercial Axiom (R) 30 K SerraSNP array of pacu. When genotype imputation was performed between densities of 9 K to 1 K, the average accuracies ranged from 0.97 (SD 0.06) to 0.90 (SD 0.12) at the SNP level and from 0.98 (SD 0.04) to 0.92 (SD 0.06) when considering the genotyped individuals. For the density of 0.5 K, low accuracy values were recorded, and less than 40 % of the SNPs showed high imputation accuracies (R2 >= 0.80). Additionally, a large proportion of low and variable imputation accuracies were observed at the extremities of chromosomes when the 0.5 K density was assessed. Based on 15,166 (91.1 %) imputed SNPs with high accuracy, a 1 K SNP array was developed and validated using the Agriseq (R) tGBS platform. The high accuracy results presented in this study can serve as a basis for selecting genotyping strategies aimed at reducing the costs associated with implementing genetic improvement programs in pacu.