Genotyping-by-sequencing (GBS): a novel, efficient and cost-effective genotyping method for cattle using next-generation sequencing

Abstract

High-throughput genotyping methods have increased the analytical power to study complex traits but high cost has remained a barrier for large scale use in animal improvement. We have adapted genotyping-by-sequencing (GBS) used in plants for genotyping 47 animals representing 7 taurine and indicine breeds of cattle from the US and Africa. Genomic DNA was digested with different enzymes, ligated to adapters containing one of 48 unique bar codes and sequenced by the Illumina HiSeq 2000. PstI was the best enzyme producing 1.4 million unique reads per animal and initially identifying a total of 63,697 SNPs. After removal of SNPs with call rates of less than 70%, 51,414 SNPs were detected throughout all autosomes with an average distance of 48.1 kb, and 1,143 SNPs on the X chromosome at an average distance of 130.3 kb, as well as 191 on unmapped contigs. If we consider only the SNPs with call rates of 90% and over, we identified 39,751 on autosomes, 850 on the X chromosome and 124 on unmapped contigs. Of these SNPs, 28,843 were not tightly linked to other SNPs. Average marker density per autosome was highly correlated with chromosome size (coefficient of correlation = -0.798, r(2) = 0.637) with higher density in smaller chromosomes. Average SNP call rate was 86.5% for all loci, with 53.0% of the loci having call rates >90% and the average minor allele frequency being 0.212. Average observed heterozygosity ranged from 0.046-0.294 among individuals, and from 0.064-0.197 among breeds, with Brangus showing the highest diversity as expected. GBS technique is novel, flexible, sufficiently high-throughput, and capable of providing acceptable marker density for genomic selection or genome-wide association studies at roughly one third of the cost of currently available genotyping technologies.

Figure 1. Distribution of the number of sequence reads and SNP call rates.

(A) number of sequence reads in individual DNA samples. (B) Call rates of SNPs (% of total SNPs called).

Figure 2. Distance between SNPs.

Distribution of the distance ranges between SNPs mapped to all bovine chromosomes for GBS and for the Illumina BovineSNP50 markers.

Figure 3. Distribution of SNPs according to chromosome size and chromosome region.

(A) Average marker density per chromosome related to its size for the SNPs from the GBS and from the Illumina BovineSNP50. (B) Average distances between adjacent GBS SNPs according to chromosomal region and length. The regional SNP distances were calculated averaging the contiguous SNPs distances in the corresponding one third of each chromosomal region. R² is the Pearson regression coefficient.

Figure 4. Detection of copy number variation (CNVs).

Read counts in all the sequenced tags and individuals for a region on Chromosome 18 that have been reported to contain a polymorphic CNV (gray area) and a SNP (yellow arrow) in the second intron of the gene SIGLEC12 that was associated with economic traits in dairy cattle . Light blue arrows show the position of the SNPs found by GBS and the dark blue are the SNPs included in the Bovine SNP50 Bead Chip. Gene annotation was obtained from MapView at the NCBI database for the bovine assembly UMD_3.1.

Figure 5. Distribution of the minor allele frequency (MAF).

MAF classes in all the animals, in Brangus cattle and in all the other breeds of cattle.

Figure 6. Neighbor-joining analysis showing the relationship among cattle breeds.

The dendrogram was constructed using TASSEL.

Figure 7. Increase in SNP numbers resulting from multiple sequencing runs.

These data correspond to SNPs detected on chromosome 1 in two lines of Sorghum bicolor (Mitchell SE, unpublished) taken from another study. Here, 384 samples per library were run in a single lane. Since the sorghum genome (730 Mbp) is approximately ¼ of the size of the cattle genome, this sorghum 384-plex is comparable to 96-samples per lane in cattle, which have an estimated genome size of 3000 Mbp .