Affymetrix, armed with a new genotyping algorithm and facing increasing competition in the whole-genome genotyping market, last week said it will reduce the price of its Mapping Array 500K set by as much as 75 percent.
However, the move, which Affy hoped would highlight growing efficiencies, was not viewed positively by investors. Shares in Affy are down more than 9 percent since the company disclosed the new price last week.
According to Affy, the new algorithm, co-developed with researchers at the University of California at Berkeley and the Broad Institute, has freed up half the space on the two 250K chips that comprise its 500K set and has enabled the company to “immediately” cut its price to $250. Depending on whether the customer is academic or non-academic, the price has been cut by between 70 percent and 75 percent.
The algorithm has also enabled the company to consolidate the two chips into a single 500K product, which is due to become available before the end of the year.
The company said it will launch a two-chip 1 million SNP genotyping set during the first quarter of 2007.
Gregory Marcus, senior product manager of DNA analysis at Affymetrix, told BioArray News’ sister publication BioInform last week that the new algorithm is based on work done on the firm’s 100K Array by Terry Speed at UC Berkeley. The algorithm, called the Bayesian Robust Linear Model with Mahalanobis distance classifier, or BRLMM, was first published in Bioinformatics in January [Bioinformatics. 2006 Jan 1;22(1):7-12].
“Affymetrix was able to eliminate the mismatch probes that were not used by the BRLMM algorithm,” Marcus said. “This freed up half of the chip real estate, which enabled us to load more content on each array.”
Other collaborators included researchers from the Broad Institute, who “did a lot of work creating algorithms for downstream analysis,” according to Marcus.
He added that no technologies are being licensed as part of the collaboration.
Illumina Sees ‘No Direct Impact’
Affy disclosed its price-cutting and new-product plans half an hour after Illumina, its chief genotyping rival, released it second-quarter financial results.
In a conference call with investors, Illumina CEO Jay Flatley said Affy’s announcement would have no effect on his company’s strategy going forward.
“We feel very comfortable with the position of our products in the market and the pricing of those products [see related story]. [link to Illumina story]
“Affy’s announcement doesn’t affect Illumina’s guidance going forward. We feel very comfortable with the position of our products in the market and the pricing of those products. We don’t see any direct impact.”
Flatley said. “We don’t see any direct impact.”
Illumina has not publicly discussed a timeline for releasing a 1-million SNP genotyping product and the company could not be reached to comment on the future of such a product.
Affy first touched upon the existence of its new algorithm in May during an investor conference when Chief Financial Officer Greg Schiffman said that Affy's original data-analysis algorithm for the 500K "wasn't fully optimized" when the company released the chips last October (see BAN 5/9/2006).
According to Schiffman, Affy’s original algorithm “didn't allow the power of the array to be fully realized and as a result we saw customers experiencing delays in scaling up to their 500K product." He added that the new algorithm would be in the hands of all Affy 500K customers by the end of May.
Affy has disclosed a series of difficulties related to the launching of the 500K set, among them a manufacturing problem that forced the company to revise its Q3 2005 guidance and a timing issue with reorders that resulted in lower-than-projected revenues for the first quarter of this year.
Schiffman pointed out in May that the problems with the company’s initial algorithm had created scale-up issues for Affy customers which, in turn, affected the timing of reorders.
"[With the older software] there was a larger degree of disparity between heterozygote calls and homozygote calls," Schiffman told the investors. "The new software algorithm takes the same information from the chip that we've already processed but enables them to extract all the information off the chip. It really has corrected the challenges we've seen in terms of the software," he said.
Marcus, the product manager, told BioInform last week that the BRLMM algorithm is delivered in three forms: via software with a user interface, a command line tool, and an open source tool. He added that the “call rate for heterozygote genotypes are now the same as the call rates for homozygote genotypes.”
“The previous algorithm tended to under-call heterozygotes, which could have introduced a bias in downstream analysis,” he said.