A risk-reduction strategy has brought Cephalon success, but now brings it face to face with a new set of risks that add up to a real management dilemma. The company is waiting to hear whether FDA will approve an expanded label for Provigil, a treatment for narcolepsy. A yes could position the product for marketing to GPs. The trouble is that four big generic companies are taking aim at Provigil. Should Cephalon spend on a big sales force, or seek a co-promotion deal that, given the situation, is bound to have tough terms? This much is clear: no one escapes risk.

Impatient with the progress of genomics, investors are looking to a new set of proteomics technologies to provide the necessary boost in research productivity. Celera, Incyte, and Myriad, the heads of which we interview in this article, have each created widely different strategies for accessing the new opportunities. Aside from differences in technology approaches, the strategies of all three firms attempt to directly address the investment world's dissatisfaction with platform models-the original business model of both Incyte and Celera. Myriad has taken a low-risk leap into the world of proteomics, spinning off its proteomics tools into a new database company that will map the proteome, delivering Myriad the information it needs for its own product discovery and development programs and selling it at the same time on a non-exclusive basis to subscribers.Celera is radically modifying its information-only strategy, transforming itself into a drug discovery company that will be assisted financially with proceeds from its database business, which will itself be boosted by new proteomic data Celera will create, review before others, and then make available to subscribers. Incyte is continuing to focus on selling information as its base business, but is adding proteomic features to its basic systems through acquisition and internal development. At the same time, it is trading access to its LifeSeq database to new proteomic technology developers for a share either of their downstream product revenues and/or rights to sell the information those products generate.

There's growing conviction that gene arrays are less useful than first imagined for expression studies and diagnostics--and that directly reading protein expression will more likely provide an accurate picture of biological status-health, disease, and pharmaceutical response. Thus, a number of companies have started up to create, on the analogy of gene arrays, protein arrays, to allow the simultaneous and quantitative testing of large numbers of proteins-potentially thousands.But the task is easier said than done. First, there are too few proteins known to allow testing for worthwhile expression patterns; and, second, because of the delicacy and variety of proteins, surface chemistry and other problems have historically made arrays impractical. None of this daunts the start-ups who, opportunistic, are looking to exploit the markets they think they can get to first: clinical diagnostics and pharmaceutical clinical development programs.

The fact that proteomic diversity is most likely several logs greater than genomic diversity is only one reason why moving from gene expression-based analyses to proteomics-based drug development presents a daunting challenge. The large issues go well beyond protein isolation and characterization: Genomics simply doesn't tell you how proteins function in vivo, and companies are just now beginning to amass and integrate the tools to correlate protein structure and function. In many ways, proteomics heralds the second coming of rational drug design, enabled now by high-throughput tools and the algorithms of computational biology. But whether structure-guided drug development will prove to be efficient or cost-effective remains to be seen.