Development and Evaluation of a Barley 50k iSelect SNP Array

Abstract

High-throughput genotyping arrays continue to be an attractive, cost-effective alternative to sequencing based approaches. We have developed a new 50k Illumina Infinium iSelect genotyping array for barley, a cereal crop species of major international importance. The majority of SNPs on the array have been extracted from variants called in exome capture data of a wide range of European barley germplasm. We used the recently published barley pseudomolecule assembly to map the exome capture data, which allowed us to generate markers with accurate physical positions and detailed gene annotation. Markers from an existing and widely used barley 9k Infinium iSelect array were carried over onto the 50k chip for backward compatibility. The array design featured 49,267 SNP markers that converted into 44,040 working assays, of which 43,461 were scorable in GenomeStudio. Of the working assays, 6,251 are from the 9k iSelect platform. We validated the SNPs by comparing the genotype calls from the new array to legacy datasets. Rates of agreement averaged 98.1 and 93.9% respectively for the legacy 9k iSelect SNP set (Comadran et al., 2012) and the exome capture SNPs. To test the utility of the 50k chip for genetic mapping, we genotyped a segregating population derived from a Golden Promise × Morex cross (Liu et al., 2014) and mapped over 14,000 SNPs to genetic positions which showed a near exact correspondence to their known physical positions. Manual adjustment of the cluster files used by the interpreting software for genotype scoring improved results substantially, but migration of cluster files between sites led to a deterioration of results, suggesting that local adjustment of cluster files is required on a site-per-site basis. Information relating to the markers on the chip is available online at https://ics.hutton.ac.uk/50k.

Keywords: SNP; barley; exome capture; genotyping chip; iSelect.

Figure 1

Principal Coordinate Analysis (PCoA) of 394 accessions genotyped with 6,917 polymorphic SNPs. The core set of 148 barley lines chosen are classified into 5 groups: (i) Central European old cultivars; (ii) spring 6 row; (iii) spring 2 row; (iv) winter 2 row and; (v) winter 6 row, and the remaining 246 accessions (classified as “rest”) are shown.

Figure 2

Plot of 9k SNP marker genetic map positions vs. physical map positions for chromosome 1H.

Figure 3

Physical distribution of the SNPs in the initial SNP set, the additional set targeted at markers in the distal chromosome regions, and the combination of these two sets (chromosome 1H).

Figure 4

Boxplot of genotype call agreement rate (%) from the comparison of calls from the exome capture variant calling and the GenomeStudio calls for the accessions genotyped with the new 50k chip. The data shown is categorised by the provenance of the SNPs (9k = SNPs from the existing 9k chip, EXCAP SNPs = SNPs from exome capture data that are new and exclusive to the 50k chip), the source and type of cluster file (JH, James Hutton; TG, TraitGenetics, adjusted vs. unadjusted) and the source of the 50k genotype call set used (JH vs. TG). Bold horizontal lines represent the median, box boundaries upper and lower quartiles, whiskers maxima and minima, and open circles represent outliers.

Figure 5

Comparison of genetic map and physical positions of SNPs on chromosome 6H mapped in GPxMo RIL population.

Figure 6

Comparison of genetic map and physical position on chromosome 5H around the position of HvDEP1, the causal gene underlying the ari-e.GP mutant phenotype mapped in the GP × Mo RIL population. SNPs from the 50k are shown as black diamonds and the three GBS SNPs are shown as red circles.