What does the microarray world look like from the summit of the mountain?
If anyone knows, it’s Affymetrix CEO Steve Fodor. Fodor didn’t invent the first microarray — that honor goes to Oxford University biochemist Edwin Southern — but, then again, Henry Ford didn’t need to invent the car to create the automobile industry. Similarly, over the past eight years, Fodor has built an industry around the GeneChip, the Model T of microarrays. He has turned a 30-person startup into a $2.4 billion company with 800 employees and customers around the world.
Now, Fodor faces a new challenge: As manufacturing giants like Corning and Motorola prepare to contest Affymetrix’s microarray dominance, Fodor needs to provide leadership that will ensure the company stays on top. But the 47 year-old Seattle native with a PhD in chemistry from Princeton is hardly losing sleep over the new competition.
“The fact that these large companies covet our space and are trying to enter this field is a huge validation of the market we have pioneered,” said Fodor.
In Fodor’s view, Affymetrix possesses not only the “first mover” advantage, but also stands to benefit from its patented photolithography process of manufacturing chips. This process, in which oligonucleotides are synthesized on GeneChips in situ by exposure to light while other parts are masked, “is the best method for producing commercial volumes of high-density DNA arrays and thus will allow us to provide a better value solution over other alternatives,” Fodor said.
“Photolithography also has the added advantage of benefiting from decades of experience and billions of dollars of investment in the semiconductor industry,” he said.
Still, Fodor knows Affymetrix also has to innovate in order to grow its business. The company fosters innovation, he said, not only by trying to hire the brightest minds as every genomics company hopes to do, but also by reaching out to scientists regularly through user meetings and informal conversations.
For example, Dimiter Dimitrov, a senior investigator at the National Cancer Institute who attended a recent Affymetrix user group meeting, said he found the meeting to be quite helpful in resolving issues he has had with data analysis — “in particular, the negative numbers for the expression levels of some genes.”
The strategy of working directly with scientists has allowed Affy- metrix of Santa Clara, Calif., to capture most of the market share for pre-fabricated arrays.
But not all researchers are happy with Affymetrix. Small labs complain it is impossible to get a response from the company when they have a problem and many scientists have commented on Affymetrix’s lack of candor when it comes to data analysis issues. Key ongoing problems researchers have identified include data normalization between different chips and random hybridization of the mismatched “control” oligos on chips.
Affymetrix has assured users it is addressing these problems, both by beefing up its internal bioinformatics capabilities and by partnering with outside companies to provide better analytical tools.
“In the new Affymetrix software that will be available later this year those mismatch probes which give signals higher than the perfect match probes will be discarded – an idea already used by Dr. M. Cam and others at NIH for their data analysis,” Dimitrov said, conveying information he learned at the user’s meeting.
In its efforts to improve, Fodor said Affymetrix is open to the possibility of acquiring another company that can “enhance our customers’ experience with our product line,” just as its 1999 acquisition of Genetic MicroSystems allowed the company to add a spotted array product line for researchers who make their own arrays.
Another issue that has plagued Affymetrix customers is the high price of Affymetrix GeneChips. Academic researchers pay about $300 per chip, while pharmaceutical companies can pay over $1,000 per chip.
Fodor defends this pricing strategy by noting that the company has decreased its “price per data point” sharply. “We’ve been able to reduce the cost per gene from $1.00 in 1994 to below ten cents in 2000,” he said.
The company plans to reduce this price-per-point further as it moves toward higher density chips. “Currently our feature size is 20 microns, but we have shown feasibility down to 2 microns,” Fodor said. “The semiconductor industry is now printing at sub 0.2 microns — obviously we have a lot of room there to shrink and provide more information to customers on a single chip.”
While Affymetrix’s ability to put higher quantities of data on chips remains virtually unquestioned, some doubts have recently arisen about the quality of the data.
In March, Affymetrix disclosed that between 25 and 60 percent of the oligos on its Murine U74 arrays had been laid down in reverse order. The data came from the public UniGene database, which contains sequence laid down in different directions (5’ to 3’ as well as 3’ to 5’), and company scientists had misread ambiguously labeled sequence data. Affymetrix immediately announced it would replace U74 sets within six weeks, and said it would also provide researchers with information on which sequence was defective.
Affymetrix’s competitors have cited this mishap as a reason for using pre-validated sequence from private companies such as Incyte instead of public data. But Fodor has taken a different lesson from the experience.
“One of the fundamental messages to customers and industry that resulted from this issue was that Affymetrix will take responsibility for the data on our chips,” he said. “If that means we need to exert a sizable amount of effort to ensure that the sequence on a chip is valid — then we’ll do it.”
Fodor added that the company has no plans to seek content from private sources. Instead, he said Affymetrix plans to continue to improve its bioinformatics capabilities for analyzing sequences and designing probes.
The company is also looking to expand its applications for its chips. Today, the chips are for research use, but with partnerships with Roche Molecular Systems, bioMerieux Vitek, and Beckman Coulter, Affymetrix is well-positioned to become a future leader in the diagnostics arena.
“Diagnostics is another important area and we believe that our platform could provide a powerful option for medical facilities with a need to analyze genetic information quickly and accurately,” said Fodor.
The company’s partnerships, which involve development of diagnostic tests with its p53 cancer array and p450 drug metabolism arrays, provide expertise with FDA approval that Affymetrix lacks, Fodor said.
Analysts have said the diagnostics area could blossom down the road, (see p. 3) but it remains speculative. In the meantime, the company plans to focus on serving its core customer base in the scientific community. “The challenge is to move the increasingly complex and ever more powerful technology into the hands of customers so they can continue to push the boundaries of scientific discovery,” Fodor said.