Thanks to high throughput screening, more than 10,000 compounds with biological activity against specific targets are entering the drug discovery process each week. But unless those compounds can pass the hurdles of bioavailability and safety, comprising a series of tests known as ADMET (for absorption, metabolism, distribution, elimination and toxicity), they will never be successful drugs. Today, the tests that make up ADMET evaluation are low throughput, and are apparently not informative or accurate enough to predict a drug's probability of success, given the high failure rate of compounds at all stages of development. Drug discovery companies are therefore looking to re-engineer the ADMET process, moving it up the early discovery chain. The goal is to predict, very early in the process, perhaps even before compounds are synthesized, which compounds pass the test for a good drug. Doing so will require new assays, new computer models, and large volumes of consistent, high quality data on drugs in man, across diverse sets of chemistries. No one company has it all; partnerships and consortiums aim to bring together the necessary resources to integrate absorption, metabolism, and toxicity into a single platform.

Coelacanth believes that its integrated drug discovery platforms will provide a solution to the preclinical bottleneck by focusing on the quality of leads rather than the quantity.

Dismal conditions in the public capital markets can't help but affect biotech start-ups too. Venture capitalists are reassessing operating and exit strategies, and contemplating new business models.

Thanks to genomics, combinatorial chemistry, and high-throughput screening, pharmaceutical companies have more drug leads than ever. But which are the drugs? Start-up firms with faster, smarter approaches to preclinical testing may help sort out the answers. Information technology will play a key role in lead optimization as new firms with simulation software are helping drug developers plot development strategies earlier than ever.