Skip to main content
Premium Trial:

Request an Annual Quote

Study Determines Best Illumina, Affymetrix Chips to Genotype European, African, Asian Populations

Premium

When it comes to arrays, researchers engaged in genome-wide association studies today have many options from which to choose. And given the advent of chips optimized for specific populations, it makes even more sense for researchers to find a way to evaluate different platforms.

The only issue has been how to accomplish that.

"Researchers always have to make a compromise between coverage and price that is proportional to the number of SNPs," said Ngoc-Thuy Ha, a PhD student at the University of Bern in Switzerland. "This is also one reason why the population-optimized SNP chips are so attractive," she said. "They promise to provide the highest coverage at a low price if your individuals are from a certain population."

Ha and fellow researchers published a study in the European Journal of Human Genetics last month that sought to answer the question of which commercial high-density SNP chip covers most of the human genome given a fixed budget. Altogether, they evaluated a dozen different Illumina and Affymetrix arrays for the study, generating assessment metrics, gaining insight into which chips are best for studying certain populations, and scrutinizing vendors’ advertised coverage claims.

Specifically, the researchers looked at Affymetrix's Axiom Genome-Wide Human EU, which is targeted to Europeans; its Axiom Genome-Wide Human ASI and Axiom Genome-Wide Human CHB arrays, which are targeted to Asians; and its Axiom Genome-Wide Human PanAFR array, targeted to Africans. It also looked at Illumina's high-density, whole-genome Human OmniExpress, Human Omni1S-8, Human Omni2.5-8, and Human Omni2.5S-8 BeadChips.

According to Ha, researchers in the past mostly relied on SNP coverage to decide which chip was most suitable for a particular association study. However, she noted, not only is SNP coverage on any given array tied to ethnicity, but also how many SNPs are on the chip to begin with. And while higher density arrays may offer higher coverage of SNPs of interest, the greater amount of content often comes with a bigger price tag, meaning that those who select such a high-density chip can genotype fewer individuals.

"This, however, will substantially decrease the statistical power to detect any association, which can really hamper the success of the study," Ha said.

Because of these issues, Ha and fellow researchers decided to look beyond coverage to assess arrays for use in GWAS. They introduced two new measures that they claim can help researchers make a final decision – efficiency and cost-benefit ratio.

To calculate both measures, they relied on the chips' sizes measured by the number of SNPs on the chips as a substitute for price "after demonstrating that there is a strong positive linear correlation between the two variables," according to the paper. Chip efficiencies were also calculated with regard to a range of linkage disequilibrium thresholds, "therefore allowing a more complete picture of the genomic coverage each chip provides."

Using these measures, in addition to coverage, they determined that the Affymetrix population-optimized arrays offered the most cost-effective coverage for the Asian and African populations. But for the European population, they established the Illumina Human Omni 2.5-8 as the preferred choice.

"Interestingly," the authors wrote in the paper, "the Affymetrix chip tailored toward an Eastern Asian subpopulation performed well for all three populations investigated." However, their coverage estimates calculated for all chips "proved much lower than those advertised by the producers," they wrote. All of the researchers' analyses were based on the 1000 Genome Project as a reference population.

According to Ha, the fact that Affymetrix has designed arrays specifically for the African and Asian populations, while Illumina has continued to concentrate on arrays with content specific to Europeans, "may be an important reason for our results."

She also said that the study should encourage researchers to use population-optimized arrays, rather than catalog chips.

"This is what I hope, since we have demonstrated that the design of SNP chips especially for the Asian and African population was indeed successful," said Ha. "It is worth having a look at the population-optimized SNP chips, before making a final decision if researchers want to keep their costs as low as possible while having the most power for their GWAS."