Development and validation of a high density SNP genotyping array for Atlantic salmon (Salmo salar)

Abstract

Background: Dense single nucleotide polymorphism (SNP) genotyping arrays provide extensive information on polymorphic variation across the genome of species of interest. Such information can be used in studies of the genetic architecture of quantitative traits and to improve the accuracy of selection in breeding programs. In Atlantic salmon (Salmo salar), these goals are currently hampered by the lack of a high-density SNP genotyping platform. Therefore, the aim of the study was to develop and test a dense Atlantic salmon SNP array.

Results: SNP discovery was performed using extensive deep sequencing of Reduced Representation (RR-Seq), Restriction site-Associated DNA (RAD-Seq) and mRNA (RNA-Seq) libraries derived from farmed and wild Atlantic salmon samples (n = 283) resulting in the discovery of > 400 K putative SNPs. An Affymetrix Axiom® myDesign Custom Array was created and tested on samples of animals of wild and farmed origin (n = 96) revealing a total of 132,033 polymorphic SNPs with high call rate, good cluster separation on the array and stable Mendelian inheritance in our sample. At least 38% of these SNPs are from transcribed genomic regions and therefore more likely to include functional variants. Linkage analysis utilising the lack of male recombination in salmonids allowed the mapping of 40,214 SNPs distributed across all 29 pairs of chromosomes, highlighting the extensive genome-wide coverage of the SNPs. An identity-by-state clustering analysis revealed that the array can clearly distinguish between fish of different origins, within and between farmed and wild populations. Finally, Y-chromosome-specific probes included on the array provide an accurate molecular genetic test for sex.

Conclusions: This manuscript describes the first high-density SNP genotyping array for Atlantic salmon. This array will be publicly available and is likely to be used as a platform for high-resolution genetics research into traits of evolutionary and economic importance in salmonids and in aquaculture breeding programs via genomic selection.

Figure 1

Source of the SNPs on the ssalar01 array. Proportion of total SNPs derived from each of the SNP discovery categories (RR-Seq, RAD-Seq, RNA-Seq and other). ‘Putative SNPs’ comprise the 286,021 putative SNPs placed on the array, and ‘QC-filtered SNPs’ comprise the 132,033 final quality-control filtered SNPs used for analysis. Note that some SNPs were detected in multiple discovery categories (see Additional file 1: Table S1).

Figure 2

Population segregation of SNPs and minor allele frequency. (A) Sharing of the QC-filtered SNPs (with minor allele frequency higher than 0.05) between the different Atlantic salmon populations depicted by a Venn diagram (number of SNPs given in parentheses). (B) Distribution of the minor allele frequency of the final QC-filtered SNPs across all unrelated animals in the test population.

Figure 3

Genomic distribution of SNPs and comparison of linkage maps. (A) Number of final QC-filtered SNPs per reference genome contig; Number of final QC-filtered SNPs contained per reference genome contig. (B) Scatterplot of number of SNPs per chromosome comparing the current study to the map of Lien et al. [3]. Note that chromosomes 2, 6, 22 and 23 are not included because the number of SNPs on those chromosomes is underestimated in the current study (see ‘Methods’).

Figure 4

Clustering of samples based on genetic similarity. Clustering of samples based on genome-wide identity-by-state and multidimensional scaling to detect population structure.

Figure 5

Use of Y-specific probes to predict phenotypic sex. Correspondence between genetic sex of the fish (based on the Y-specific probes on the array) and phenotypic sex (where known).