Skip to main content
Premium Trial:

Request an Annual Quote

As Cost of SNP Genotyping Decreases, Affy Advocates Change in Study Design

Premium
It is no secret that Affymetrix’s unofficial flagship product is its Genome-Wide Human SNP Array 6.0, called the 6.0 for short, which helped buoy the firm’s second-quarter revenues by 10 percent year over year after only being on the market for part of the quarter.
 
But while Affy has been touting the 1.8-million marker array — which contains both SNP and copy number variation content — as the best chip in the market, it has also been strongly advocating that scientists using the arrays in whole-genome association studies take advantage of their relatively low cost to replicate their finding using the same array, rather than validating their SNPs of interest via custom genotyping methods.
 
Affy first slashed its genotyping prices in the summer of 2006 to regain a foothold in the market while facing intense competition from Illumina, which is the only other company with the resources to challenge Affy in the whole-genome genotyping array space (see BAN 7/25/2006).
 
Specifically, the company lowered the price of is 500K Mapping Array Set to $250 per chip, which was followed by the launch of its 1-million marker SNP 5.0 Array in January, and now the SNP 6.0, which sells for around $400. According to Affymetrix President Kevin King, the price factor could change the way researchers approach association studies and could ultimately benefit Affy’s top-selling product.
 
“It provides a cost-effective price point for larger studies and it is allowing users to do replication studies instead of targeted genotyping,” King told investors during UBS Global Healthcare Services Conference in New York last week.
 
“In the past, because of whole-genome association cost, customers typically divided their experiments into two pieces,” he said. “First they did an initial whole-genome association study, followed by a replication phase with a smaller subset of SNPs,” King said.
 
“What has changed with 6.0 with our higher content and compelling economics is that now customers can cost-effectively interrogate the entire population at both the whole-genome level as well as at the replication level with 6.0,” he added.
 
Winning Converts
 
Because it takes time to design a genome-wide association study, there are relatively few papers available that detail the benefits of Affy’s recommended study design. During his presentation at UBS, King cited a recent paper in the New England Journal of Medicine that used Affy’s arrays to identify more markers during the replication phase than during the first phase [Samani N, et al. Genome-wide association analysis of coronary artery disease. New England Journal of Medicine. 2007 Aug 2;357(5):443-53.]
 
Some Affy users are already designing experiments that include a replication phase using the 6.0, as opposed to a custom genotyping study as they would have in the past. According to Alexandre Stewart, a principal investigator at the University of Ottawa Heart Institute, which is performing association studies on coronary artery disease, both pricing and timing made replication a compelling option in study design.
 
“When we initially considered customizing, we were using 500K arrays. But the 6.0 arrays have such higher coverage that they have made the need for customizing somewhat obsolete,” he told BioArray News last month. “Cost is a factor. You can do it more cheaply with a 6.0 array than with a customized array,” he said.
 
Stewart said that the concept of a replication study is not a new one, but that the previous costs associated with genotyping had been prohibitively high. “No one could afford to do replication studies in the past. In fact, when we started, we couldn’t afford to do the Affy chips,” he said.
 
“When we started out this study with the 500K chips they were CAN$1,500 ($1,501) a pair per patient,” he said. “So if you run a thousand patients and a thousand controls, that becomes an astronomically expensive venture, but we took a chance and we figured the price would come down if the competition came in and that is exactly what happened,” he added.
 
Stewart said that the Ottawa Heart Institute is now moving ahead with plans to replicate its original work with the 500K in phase II of its study on 3,000 samples this year, and then another 12,000 samples next year.
 
Another Affy user, Ulrich Broeckel, the associate director of the Children's Research Institute at the Medical College of Wisconsin, is also considering the 6.0 for use in a replication study. Broeckel told BioArray News in an e-mail last month that in the past, CRI “selected significant markers and typed [them] with a different platform such as TaqMan,” but now the institute will use the 6.0 instead.
 
“The density of the 6.0 is obviously higher compared to previous versions [of Affy genotyping arrays], so this should reduce the number of SNPs which will be required for replication,” he wrote.
 
A Context (and Platform)-Dependent Decision
 
However, Broeckel also wrote that there is still a need for custom genotyping despite the potential of using the 6.0 in a replication study. “I still see the need for replication with an even higher-density set of SNPs; more SNPs in regions of interest where the 6.0 does not provide sufficient density to identify causal SNPs or SNPs in strong linkage disequilibrium,” he wrote.  
 

“At this point it’s really a mathematical exercise as to what gives you more power and what works for your budget.”

Broeckel’s opinion matched that of Yik Teo, an investigator at the Wellcome Trust Center for Human Genetics in Oxford, UK. Teo told BioArray News in an e-mail this week that Affy’s advocacy of replication studies makes sense for larger projects, but that other technologies and custom arrays might still be better for some researchers.
 
“These products do indeed result in lower cost per SNP when compared to designing large-scale
custom arrays of a few thousand SNPs,” he wrote. “However, for replicating small numbers of putative associated SNPs — of the order of, say, less than a thousand — across thousands of additional samples, there are existing technologies which cost less and generally require a shorter time to actually obtain the genotypes for analysis,” he added.
 
Dietrich Stephan, head of the Neurobehavioral Research Unit at the Translational Genomic Research Institute, told BioArray News this week that he expects replication studies to increase in number, but that smaller studies would probably benefit from existing technologies, such as Sequenom’s MassArray platform.
 
“It is still context dependent,” he said. “If you have the high-throughput infrastructure, you can use that replication approach but it takes longer and it is probably more expensive as opposed to blasting through 10 SNPs on [the] Sequenom platform or something,” said Stephan.
 
And Affy’s approach is certainly not the consensus among array platform vendors. Unlike Affy, Illumina has not been advocating replication studies for its SNP chips. Rather, the company recommends that researchers use a high-density array, the like Human 1M BeadChip, for the first phase, and then use its iSelect custom array development tool to order targeted arrays for phase II genotyping.
 
According to Carsten Rosenow, marketing manager for DNA analysis solutions at Illumina, the company has experienced “tremendous growth in custom genotyping.” Rosenow told BioArray News last week that the “two-phase project in whole-genome association study design is still very popular” and that there is a “huge interest” in Illumina’s iSelect custom program.
 
He noted that because Illumina’s custom arrays are manufactured on 12-sample chips, they enable users to reduce the cost and time of their whole-genome association studies. Still, Rosenow said that Illumina is ready to support those who wish to do replication studies, rather than custom genotyping.
 
“At this point it’s really a mathematical exercise as to what gives you more power and what works for your budget,” he said. “I think we are in a unique position because we can offer solutions for both areas.”