Despite the challenges of neuropsychiatric drug discovery, focused research into CNS pathways has raised hopes that R&D can be rationalized sufficiently to lower risk. A new generation of biotechs is about understanding the neural mechanism of disorders and then finding compounds that address the mechanisms.

Until now, Perlegen contends, pharmacogenomics has failed to deliver because no one has had the statistical power to show connections. With its whole genome scanning technology, Perlegen aims to prove that it can come up with usable and predictive markers of drug response by inexpensively screening for millions of SNPs and analyzing subsequent patterns--even without any significant biological understanding. The start-up aims to use its technology to help compounds win regulatory approval and more market share than they otherwise could. Perlegen also plans to support in-licensed drug candidates and seek disease genes. It wants a share of drug sales its studies enable, and to leverage all the IP it can capture. Absent its own drug discovery and development program, Perlegen needs partners to provide the compounds and clinical samples with which it can create pharmacogenomic data and secure an FDA imprimatur. Competitors argue that smaller sets of data, informed by a clearer biological understanding, will provide equally effective pharmacogenomic data, cheaper.

Pharmacogenomics has disappointed advocates who saw the opportunity to apply a discovery tool to the near-term goal of increasing approval chances and marketability for late-stage and marketed compounds. In return, they hoped to take a percentage of the highest-cost segment of the pharmaceutical budget. But Big Pharma is by and large not using pharmacogenomics for late-stage and marketed compounds: senior executives don't believe there's enough evidence it works and are afraid of limiting the marketability of the products by segmenting broad target populations into niches. Some also worry about uncovering potential side-effects that non-pharmacogenomic trials wouldn't reveal. Nonetheless, pharmacogenomics has made it to Big Pharma: most companies, for example, are banking samples from clinical trials to be pharmacogenomically tested retrospectively, thereby informing future trials. Not that this means the pharmacogenomics specialists will be able to sign high-value deals with the commercial side of drug companies, who believe that pharmacogenomic analysis is available from a number of sources, including internal ones, and feel they own the key assets for creating meaningful programs: compounds and patient samples. Instead, pharmacogenomics will find its place first as a discovery technology, integrated with other methods for finding, validating and prioritizing targets. That means that to succeed selling pharmacogenomics, biotechs will have to combine their pharmacogenomic assets with other discovery technologies, perhaps through mergers. An alternative: use their technologies to find drug products that they can themselves develop, perhaps later out-licensing them.