This year, forget Mount Everest. Any adventurer need look no further than the expanding world of microarrays, where opportunities abound to scale the heights of technological innovation, beat a new path though the genomic and proteomic wilderness, and face the most harrowing challenge (the one that sent many a dot-com flailing to its death), that of transforming hot technology into cool revenue.
As the year begins, scores of expeditions have already been launched, leader Affymetrix has begun to scale mount profit, and undaunted new teams are forming at base camp, (see chart pp. 6-7) PowerPointing toward the summits they’ll attempt, and preparing their equipment.
Among the expeditions in progress, there are those of beaded array companies such as Illumina, Lynx, and Luminex. These three have all been at it for a few years, and all have been burning cash. Illumina and Lynx are both tacking their lofty business plans on their fancy technology, while Luminex has played the expedition supplier, offering its beads to any company with a viable plan to develop them into a real research or diagnostic tool. The latter strategy is a safer bet, since one of Luminex’s 30-plus partners is bound to develop something useful and marketable. Illumina and Lynx, however, have been garnering revenue from providing genomic services, which is not necessarily a viable stand-alone strategy in that it relies on a few large make-or-break pharma customers, who can be fickle. (Look at genomic service provider Genometrix and Incyte, which both got lost in the woods and gave up their genomic services expeditions this year.)
Behind this cluster, a couple of startups offering “quantum bead” technology, implanted fluorescent particles used as barcodes in beads — are just leaving base camp. The beads are attached to oligo or peptide probes, then read by detection instruments. Quantum Dot of Hayward, Calif. is laying down a number of roads toward commercializing the fluorescent particles, but will soon have to face competition from Pittsburgh’s Launchcyte, which is incubating a startup with a bead-based system. On top of this competition, the inherent complexity of the system could forestall its rapid adoption in the microarray world.
Several other groups starting up the mountain include two competing companies, PamGene and Metrigenix, that offer Swiss cheese flow-through chips in which the probes are attached to the insides of the holes or channels in the chips; Combimatrix, a subsidiary of Acacia Research,that has piggybacked on to Roche diagnostics in its effort to make 1,000-probe semiconductor microarrays; A gaggle of universal microarray developers hoping to make arrays that can be used to probe any species using tags attached to the target molecule and complementary tags attached to the array; and two companies, febit and Nimblegen, developing machines for do-it-yourself light-based in situ synthesis of oligo arrays.
While there is plenty of territory to scale — the microarray market has been projected as reaching two to three billion by 2004 — it’s clear that many of these companies will not make it. Since many have promising technologies, and all have venture capitalists backing them, the true test will whether a company can transform its technology into a viable, user-friendly product.
Additionally, there are those that hope to scale new heights in high-density arrays such as Xanthon of North Carolina, which promises to put up to 10 million electrode probes on a square centimeter and Xeotron of Houston, which has Texas-sized ambitions to quickly make high-density in situ custom oligo chips.
But the objective these startups hope to reach — a bigger, better, faster gene expression array — is not the only challenge on the horizon. The sector has been vexed by the inadequacy of software for analyzing array data, a need that has been met with a flood of new array analysis packages. But most of these packages merely offer newer and fancy ways to cluster the data, a form of exploratory analysis that statisticians have been saying is inadequate and fails to address basic problems such as normalizing gene expression levels across chips. This means the field is still open to any company that provides statistical software for microarrays.
Another area of mounting importance is protein chips, where Ciphergen has sought to become the Affymetrix, but a number of others from Prolinx and LumiCyte to Gyros, the Swedish maker of round microfluidic microarrays, have pitched their tents. The two main issues are surface chemistry, where companies have been developing surfaces prevent non-specific binding of proteins, and antibodies, which are unlike genes, in limited supply. There is also the issue of which proteins to spot down, — a question that makers like Prolinx, Packard Bioscience, and Biosite have avoided by producing do-it-yourself chips without probes.
Amid all these vaunted plans, there is still one basic issue to resolve: standardization. The problem of exchanging information across platforms will even grow in importance if the widely differing types of arrays now being developed become commonly used. The Microarray Gene Expression Database working group will attempt to address these issues of standardization as it convenes February13 to 16 in Boston.
Even with standards on the horizon, there will be new challenges to address. One of these is the phenomenon of alternative splicing, wherein different permutations of exons in a gene get spliced together to make different proteins or other molecules. Learning about the splice variants of a gene may help to assign ESTs to genes, and aid in the interpretation of microarray data. Also in the January issue of Nature Genetics, UCLA biochemists Barmack Modreck and Christopher Lee discuss the potential of microarrays to detect splice variants. They suggest that long probes of 60 nucleotides could be designed to span exon-exon junctions. “As alternative splicing of a given gene creates different exon-exon junctions, it can be detected by measuring hybridization of mRNA samples from different tissues to these probes,” they write in the article,A genomic view of alternative splicing.
Who knows, maybe splice-variant arrays will join SNP arrays and protein arrays as the newest applications of this evolving technology. The adventure is just beginning.