Technical note : High fidelity of whole-genome amplified sheep ( Ovis aries ) deoxyribonucleic acid using a high-density single nucleotide polymorphism array-based genotyping platform 1

Advances in high-throughput genotyping technologies have afforded researchers the opportunity to study ever-increasing numbers of SNP in animal genomes. However, many studies encounter difficulties in obtaining sufficient quantities of highquality DNA for such analyses, particularly when the source biological material is limited or degraded. The recent development of in vitro whole-genome amplification approaches has permitted researchers to circumvent these challenges by increasing the amount of usable DNA in normally small-quantity samples. Here, we assess the performance of whole-genome amplification products generated from ovine genomic DNA using a high-throughput SNP genotyping platform, the newly developed Illumina ovineSNP50 BeadChip. Our results demonstrate a high genotype call rate for conventional genomic DNA and whole-genome amplified genomic DNA. The data also reveal an exceptionally high concordance rate (≥99%) between the genotypes generated from whole-genome amplified products and their conventional genomic DNA counterparts. This study supports the use of whole-genome amplification as a viable solution for the analysis of high-density SNP genotypic data using compromised or limited starting material.


INTRODUCTION
Single nucleotide polymorphisms are the most common form of DNA sequence variation in mammalian genomes (Kruglyak and Nickerson, 2001).The abundance and pan-genomic distribution of SNP have resulted in their adoption as the marker of choice for mammalian genome-wide association studies.These surveys have been enabled by the availability of high-throughput genotyping platforms, such as genotyping arrays (God-dard and Hayes, 2009).Although genotyping arrays have greatly reduced the amount of DNA required for large-scale genotyping, insufficient quantities of genetic material can hamper investigations, especially when the source biological material is limited or of poor quality or both (e.g., forensic, ancient, and archival samples).To facilitate the analysis of compromised samples, wholegenome amplification (WGA) technologies have been developed.Indeed, one such WGA method, multiple displacement amplification (MDA), has been particularly successful, generating high-quantity, high-fidelity, and uniformly amplified whole genomic DNA (Lovmar and Syvanen, 2006).
Previous studies assessing the performance of WGA-DNA using SNP genotyping arrays have demonstrated ≥98% concordance in genotype calls between nonamplified, conventionally treated genomic DNA and WGA-DNA products generated from the same samples  (Croft et al., 2008;Jasmine et al., 2008;Xing et al., 2008).Here, we investigate the performance of WGA sheep DNA using the recently developed Illumina ovine SNP50 BeadChip.To our knowledge, this study represents the first assessment of the performance of WGA sheep DNA using a high-density genotyping platform.

MATERIALS AND METHODS
This work has been approved by the University College Dublin, Ireland, Animal Research Ethics Committee.
The DNA from 5 unrelated Irish Suffolk sheep (4 males and 1 female) was analyzed in this study.The MDA-WGA was performed on all 5 samples using the REPLI-g Midi kit (Qiagen, Crawley, UK) and purified conventional genomic DNA as template according to the manufacturer's instructions (Table 1 and supplemental information; http://jas.fass.org/content/vol88/issue10/).Conventional genomic DNA (herein termed genomic DNA) and WGA-DNA samples (labeled gDNA1-5 and WGA-DNA1-5, respectively) were subsequently analyzed using the Illumina ovineSNP50 BeadChip (Illumina, San Diego, CA), which assays a total of 49,034 evenly spaced SNP loci distributed across the ovine genome.Raw signal intensities were converted into genotype calls using the Genome Studio software package (version 2008.1,Illumina) and a cluster file derived from approximately 3,400 sheep representing more than 70 breeds that form the International Sheep Genomics Consortium's HapMap sample set (http://www.sheephapmap.org).

RESULTS AND DISCUSSION
The performance of the genomic DNA and WGA-DNA samples on the ovineSNP50 BeadChip are summarized in Table 1.The WGA-DNA products yielded between 17 to 23 µg of DNA, corresponding to a 49-to 124-fold increase in yield.The sex of each animal was also correctly inferred from the resulting SNP genotypes from both sample preparations.
To calculate the SNP genotype call rate for all samples, we counted the total number of loci that failed to yield a genotype (i.e., no-call SNP) in the genomic DNA and WGA-DNA products.Across all 10 arrays: 1,624 SNP genotypes from 1,227 different SNP loci were classed as no-call; 1,507 genotypes were classed as no-call in the WGA-DNA samples only; 77 in the genomic DNA samples only; and 20 in genomic DNA and WGA-DNA.Notably, no single SNP locus failed across all 10 assayed DNA samples.The genotype call rates for the 5 genomic DNA samples ranged between 99.84 and 100% [mean call rate of 99.96%, SD of 0.07%].The mean call rate for the WGA-DNA products was less than the genomic DNA samples, ranging between 98.93 and 99.87% (mean call rate = 99.38%,SD = 0.45%).Overall, the mean call rate for the WGA-DNA products was 0.58% less than the genomic DNA samples.The SNP genotype call rate observed in our study is comparable with that reported for previous studies involving WGA-DNA, which ranged between 92.98 and 100% for genomic DNA-WGA-DNA sample pairs (Croft et al., 2008;Jasmine et al., 2008;Xing et al., 2008).
We next investigated possible SNP allele amplification bias induced by the WGA technique used.For this, we selected SNP loci that failed to genotype in WGA-DNA products but were called in the corresponding genomic DNA samples.We found that the proportion of each genotype (AA, AB, BB, where A and B represent the major and minor alleles at each locus, respectively) that failed in the WGA-DNA differed from the proportion of each genotype called for the genomic DNA samples (χ 2 test, P < 10 −4 ).A disproportionately large number of SNP classed as no-call in the WGA samples were heterozygous in the genomic DNA samples (2.3× the expected proportion based on the genotypes observed in the genomic DNA samples; Table 1).These findings differ from those of Xing et al. (2008), which show an underrepresentation of AA and AB genotype calls and an overrepresentation of BB genotype calls for SNP loci that failed in WGA-DNA products.It is possible that the discrepancies between our data and those of Xing et al. (2008) reflect differences in the SNP genotyping technology used (Illumina BeadChip vs. Affymetrix SNP array), the SNP genotype imputation algorithms used [Illumina GenomeStudio vs. the Affymetrix Dynamic Model algorithm (Affymetrix, Santa Clara, CA)], or a combination of both.
We next investigated the genotype concordance rate between the genomic DNA and WGA-DNA products from the same sheep sample.For this analysis, only those SNP that generated genotypes in genomic DNA and WGA-DNA products were considered.Discordant SNP were classified as those SNP that were (1) heterozygous (AB) in 1 DNA preparation but homozygous for either allele in the other DNA preparation (AA or BB), or (2) homozygous for 1 allele in 1 preparation and homozygous for the alternative allele in the other DNA preparation (AA→BB or BB→AA).In total, 36 discordant SNP genotypes were detected across all 10 genotyping arrays, with 32 SNP genotypes represented by single discordant pairs of samples and 2 SNP genotypes showing discordance in 2 sample pairs.In 31 instances, a SNP locus was detected as being heterozygous in a genomic DNA preparation and homozygous in its WGA-DNA counterpart, and in 5 cases as heterozygous in WGA-DNA and homozygous in genomic DNA.No discordant SNP loci displaying different homozygous calls were detected between the 2 DNA preparations from the same sample.Notably, the WGA-DNA samples display excellent concordance with their genomic DNA counterparts, ranging between 100 and 99.97% (mean concordance rate = 99.99%,SD 0.01%).These concordance rates are similar to previous investigations assessing the performance of high-throughput genotyping analyses of WGA-DNA (Table 2).Single nucleotide polymorphism genotyping Finally, we investigated the potential of systematic SNP genotype calling failure in the WGA-DNA products.Of the total 1,157 no-call SNP loci detected in any of the 5 WGA-DNA products, none failed consistently in all 5 array pairs.To analyze further the spatial distribution of both of the no-call (n = 1,227) and discordant SNP loci (n = 34), we plotted the total number of each in 2-Mb nonoverlapping windows across each ovine chromosome.These no-call and discordant loci (n = 1,261) were distributed across the genome with no apparent clustering, indicating no systematic errors with the array design or the differentially prepared DNA samples (Figure S1, supplemental information; http://jas.fass.org/content/vol88/issue10/).This finding is consistent with the observations of Xing et al. (2008).
A subsequent study performed by us compared the genotyping performance of 11 compromised WGA-DNA samples with 28 good quality, non-WGA, conventionally treated genomic DNA samples from different individuals belonging to the same pedigree on the Illumina ovineSNP50 BeadChip.Compromised samples had been stored as whole blood at 4°C for a period of 6 mo, then at −20°C for 17 mo before whole genomic DNA extraction using the methods outlined in the supplemental information.The compromised samples showed no increased allelic drop-out as evidenced by Mendelian segregation errors (compromised WGA-DNA mean 0.02%, SD 0.03% vs. genomic DNA mean 0.02%, SD 0.02%) and heterozygosity (compromised WGA-DNA mean 30.76%,SD 2.14% vs. genomic DNA mean 31.79%,SD 2.28%).The compromised WGA-DNA samples did show a reduced call rate compared with the genomic DNA samples, but the difference was not significant (compromised WGA-DNA mean 2.00%, SD 2.16% vs. genomic DNA mean 0.82%, SD 0.75%).
In summary, this study demonstrates that the newly developed Illumina ovineSNP50 BeadChip is suitable for the analysis of conventional genomic and WGA sheep DNA samples.This array produces increased SNP genotype call rates and highly concordant SNP genotypes for these DNA sample preparations.Furthermore, this study further supports the use of WGA sheep DNA as a potential viable solution for researchers who require large-scale genotype information from sheep DNA extracted using compromised or limited starting genetic material from sources such as ear punches from tags, hair or wool follicles, saliva, urine, and milk.

LITERATURE CITED
(Illumina, San Diego, CA).A and B represent the major and minor alleles at each locus, respectively.

Table 1 .
The DNA sample information and performance on the Illumina ovineSNP50 BeadChip 1

Table 2 .
The SNP genotype concordance rate for conventional genomic DNA and whole-genome amplified-DNA (WGA-DNA) samples using high-density genotyping arrays