The development, implementation and improvement of applied breeding programs in aquaculture is an area of intense activity. The Australian farmed barramundi sector now has sufficient scale to justify the development of family-based breeding. We describe the development of a SNP tool formatted for routine commercial DNA profiling and test it in terms of i) informativeness and allele frequency distribution, ii) ability to detect genetic relatedness and iii) assigning parentage in the progeny of group spawning events. To begin development of the SNP tool we performed whole genome resequencing using a discovery panel of 10 Australian barramundi. The resulting Illumina short read data set was trimmed and mapped against the available 670 Mb reference assembly (Vij et al 2016). First pass variant calling identified a total of 3,731,369 raw variants. These were subsequently subjected to a series of quality filters to identify a set of 39,873 high confidence loci. Given the discovery panel consisted of individuals from a single population, published sequence data from an additional 17 wild caught barramundi from Australia and South East Asia was downloaded (Vij et al 2016) and SNP calling performed using the same parameters. Intersect analysis revealed 83% of SNP detected in the Australian population were independently called in the set of 17 wild caught fish (33,135 / 39,873 SNP). A key objective was to develop a low-density SNP genotyping tool that could be cost effectively deployed in applied breeding. Additional prioritisation steps then identified SNP that were located within the coding genes (1,637 SNP) before a final set of 337 SNP were formatted for analysis using kompetitive allele-specific PCR (KASP) at the Center for Aquaculture Technologies (CAT, San Diego California).
A set of 771 individuals were genotyped, resulting in high rates of both SNP (93%) and samples (97%) passing quality filtering based on minimum call rates and genotype frequency testing. Genotypes from samples processed in multiple batches were compared to estimate genotyping accuracy. A total of 1107 comparisons were evaluated, resulting in an estimated genotyping error rate of 0.18% (2/1107). To investigate levels of informativeness, estimation of minor allele frequency (MAF) revealed the majority of loci (215/282 or 76%) had high MAF (> 0.30), confirming the SNP tool can be expected to have high utility for applications in applied breeding.
We tested the SNP tool’s ability to distinguish between different types of pairwise relationships using RXY_LR (Pew et al 2015). Full-sibs and parent-offspring pairs were unambiguously distinguished from unrelated animals, while the majority of half-sib pairs were convincingly distinguished from unrelated animals. We next tested the tools ability to detect population substructure. Clear blocks of elevated allele sharing (AS) were evident, and comparison with the results of formal parentage revealed AS blocks identified separate full-sib families. Taken together, the results confirmed a SNP tool containing only a few hundred loci is sufficient to support applied breeding applications in group spawning species such as barramundi.