Development and characterization of a high density SNP genotyping assay for cattle

Abstract

The success of genome-wide association (GWA) studies for the detection of sequence variation affecting complex traits in human has spurred interest in the use of large-scale high-density single nucleotide polymorphism (SNP) genotyping for the identification of quantitative trait loci (QTL) and for marker-assisted selection in model and agricultural species. A cost-effective and efficient approach for the development of a custom genotyping assay interrogating 54,001 SNP loci to support GWA applications in cattle is described. A novel algorithm for achieving a compressed inter-marker interval distribution proved remarkably successful, with median interval of 37 kb and maximum predicted gap of <350 kb. The assay was tested on a panel of 576 animals from 21 cattle breeds and six outgroup species and revealed that from 39,765 to 46,492 SNP are polymorphic within individual breeds (average minor allele frequency (MAF) ranging from 0.24 to 0.27). The assay also identified 79 putative copy number variants in cattle. Utility for GWA was demonstrated by localizing known variation for coat color and the presence/absence of horns to their correct genomic locations. The combination of SNP selection and the novel spacing algorithm allows an efficient approach for the development of high-density genotyping platforms in species having full or even moderate quality draft sequence. Aspects of the approach can be exploited in species which lack an available genome sequence. The BovineSNP50 assay described here is commercially available from Illumina and provides a robust platform for mapping disease genes and QTL in cattle.

Figure 1. The BovineSNP50 assay has a compact gap distribution ideal for genome wide association studies as compared to the Affymetrix 25 K SNP panel that has an excess of adjacent markers either too close or too far apart, leaving large sections of the genome unrepresented on the assay.

Figure 2. Distribution of SNP by call rates on the BovineSNP50 assay.

The overall call rate for all markers exceeded 99.1% and more than 90% had call rates above 99.98%.

Figure 3. Average MAF by SNP source (see Methods) demonstrates the utility of the assay in taurine, composite, African and indicine cattle.

Figure 4. Distribution of SNP minor allele frequency by SNP source.

Figure 5. Distribution of SNP MAF by group.

A – waves 1, 4, 5 and 6 with High MAF, B – waves 2 and 7 with Low MAF, C – waves 3, 8, 9 and 10 with no MAF available, and D – waves 11, 12, 13 and 14 comprising Draft SNP. (Note: Trend lines drawn only for better illustration).

Figure 6. Genome-wide association analyses for (a) coat color based on Fisher's exact test applied to allele frequencies and (b) the POLL locus genotypes based upon a likelihood ratio test for the extent of linkage disequilibrium (r2) between each SNP and the POLL locus.

Figure 7. Schema used to produce a weighting factor for assay selection for each candidate SNP depending upon location within a chromosomal interval.