Illumina’s next generation of microarrays for whole-genome genotyping is more likely to be driven by the addition of unique content drawn from second-generation sequencing projects rather than a substantial increase in the amount of content per chip, according to CEO Jay Flatley.
Flatley said last week that the company sees limited benefit to introducing arrays that are more dense than the 1.2-million-marker Infinium HD Human 1M-Duo BeadChip, which it began shipping during the second quarter. Instead, future products are likely to be introduced on the basis of content or new, multi-sample formats.
“Our current flagship array in the genotyping line has about 1.2 million variants on it,” said Flatley. “We think that that is sufficient for almost all studies that scientists want to do today.”
Flatley said that there “may be some push over the longer run to go up to 2 million markers or so,” but that Illumina’s current chips capture almost 98 percent of known variation in Caucasian populations, and slightly less in African and Asian populations.
“The incremental return for the next million SNPs is pretty low,” he said.
Flatley made his comments at Leerink Swan’s Emerging Products and Applications in Life Science Tools Roundtable Conference, held in New York last week.
For the past three years, the array-based, whole-genome SNP genotyping market has largely been driven by the amount of content the two main players, Illumina and Affymetrix, could put on a chip.
For instance, Illumina launched its 100,000-marker BeadChip in 2004 (see BAN 4/27/2004
), which Affymetrix followed the next year with its 500K Human Mapping Set, and then Illumina countered in 2006 with its HumanHap500 BeadChip.
Both companies began selling 1-million-plus marker chips in 2007 (see BAN 7/24/2007
According to Flatley, the amount of content Illumina deposits on its arrays is now likely to remain stable as the company reevaluates the data on the current Human 1M-Duo BeadChip for future product launches.
“We think the bulk of the market is going to be in the range of 500,000 to 1 million markers,” Flatley said at the conference. “What will happen over time is that the content will improve and we will learn more about which SNPs are irrelevant [and] which ones aren’t. We will throw out some of the ones that are not that valuable.”
Flatley added that Illumina will be able to “rev the content as we sequence more humans; we’ll be able to understand more about rare variation and feed those rare variants back onto the arrays.”
“There is no trend of genotyping projects migrating to sequencing.”
Illumina spokesperson Peter Fromen told BioArray News
last month that the company has already been using its second-generation sequencing platform, the Genome Analyzer II, to generate content for array products. He cited as one such product Illumina’s Bovine SNP50 BeadChip, which features 54,000 SNPs equally spaced across the cow genome (see BAN 7/29/2008).
Flatley said Illumina over the coming year will also try to stick to its strategy of product proliferation and selling arrays in multi-sample formats. Illumina’s latest generation of 2.3-million marker HD chips, the HD 610-Quad, which offers four arrays on a chip, and HD 1M-Duo, which offers two arrays, are both examples of the firm’s approach to the market (see BAN 1/8/2008
), he said.
“Clearly, multi-sample is a very important feature for our customers because it reduces their cost,” said Flatley, explaining that it “reduces our cost of manufacture and we can pass that onto the customers in terms of better pricing.
“In our flagship product we actually have two, 1.2-million arrays on that chip, so it improves the efficiency of the labs in terms of running time, and it allows us to make the chips more economically,” he added.
Genotyping Stays ‘Healthy’
At the conference, Flatley discussed how the company’s two most lucrative technology platforms, arrays and sequencers, fit the market on an application-by-application basis.
According to Flatley, Illumina views the Genome Analyzer as a better tool for several applications, namely chromatin immunoprecipitation analysis and gene expression while he said the company’s arrays are more cost-effective for genotyping and the emerging consumer genetics space.
Illumina’s principal offering in the ChIP market is its ChIP-Seq application, which competes against array-based methods offered by nearly all of its competitors, chiefly Affy, Agilent Technologies, and Roche NimbleGen.
“This is an application that studies where proteins bind to the DNA molecule in the cell, and the advantage of sequencing over arrays is that you have to make on an array an assumption of range,” he said. “When you run the analysis you get an analog output that tells you the neighborhood of where the proteins are binding to the DNA.”
Flatley said that, in sequencing, “because you are looking at every base, you can determine exactly the range of binding, and you can do this in a massively parallel way, and you do it in a digital way, so you know exactly how much on a quantitative basis has stuck.”
The company has also forecast that customers will eventually migrate from array-based gene-expression applications over to the digital gene-expression application offered on the Genome Analyzer. However, Flatley said that it was unlikely that the Analyzer would supplant arrays in genotyping or related opportunities, like consumer genetic testing.
“There is no trend of genotyping projects migrating to sequencing,” he said. “We think the genotyping market is very healthy; they are very complementary approaches, and we have not seen any impact of researchers going over or funding diverted from genotyping to sequencing.”
Illumina provides arrays to 23andMe, a Mountain View, Calif.-based firm that began offering personal genotyping last year (see BAN 7/22/2008
). “We think consumer genotyping will certainly be a big market over the next three to five years,” Flatley predicted last week, “but consumer sequencing will have to be a rich and famous market; there will be people who want to get sequenced, but it won’t be a [similar] opportunity for probably the next 10 years.” He did not elaborate.
In Flatley’s opinion, “sequencing is inherently a better tool” than analog arrays for applications like expression and protein binding.
However, genotyping is better suited to arrays because it’s “already digital,” he said. “When we run a genotyping chip, we call an A, C, G, or T, and as long as you have high confidence in that call, essentially you are getting a digital answer already.”
Therefore, Flatley predicted that “genotyping is unlikely to be done on sequencers at least at the whole-genome level for a long time as a replacement technology.” Instead, he said the technology may be used in validation studies in the future to sequence regions of interest identified during whole-genome genotyping studies.
“That is one area where there will be some kind of overlap, but in terms of supplying technologies, we are sort of agnostic as to how that gets done,” Flatley said. He added that copy number variation studies will also eventually move over to sequencing, but that chips will remain the most cost-effective way of surveying CNVs “for a couple of years.”