On the other hand, pitching the DRS program to large companies has been challenging. "Often, one of the hardest things we do is find the right person within the company to talk to," says Smith-Farrell. Frequently, no one is in charge of the molecules they want. "There is no vice president for discontinued compounds," says Gessler. No VP of dead drugs.

Gessler, who declines to speak in detail about the structure and value of specific deals, does like to recount his favorite argument for selling the repositioning program to Big Pharma. He asks his potential business partners what a particular compound would be worth if they out-licensed it "as is." If the compound had been discontinued in Phase II for lack of efficacy, for instance, everyone at the table knows it is all but worthless. Then he asks what the company would expect to pay if it had to in-license the compound once proof-of-concept was established in an animal model for another indication. "It tends to make the point about value creation," Gessler says.

Roche, in contrast to many companies, has a clear-cut policy on who controls discontinued drugs. Drugs that fail out of the clinic become the property of R&D. But even so, tracking the discontinued drugs proves challenging. "Digging up the history of these compounds within the company is not trivial," Babbis says. "And also making sure that we had sufficient compound on hand so that we don't have to synthesize the compound. We don't want this to divert from our main goal of supporting the pipeline, so we were pleasantly surprised to see that we are able to do this."

The biggest upside to the Gene Logic approach may be the opportunity to search for new indications earlier. As Babbis puts it: "Finding new indications is part of what we call lifecycle management of any compound. Probably the greatest way that you can create value is to find multiple indications for one drug or one drug series."

Instead of a serendipitous element in late-stage clinical trials or post-approval lifecycle management, the search for new indications could become a standard element of preclinical research. And although the pipeline has too few late-stage candidates, it has never been so full of preclinical compounds, many of which need not become dead drugs before developers seek unexpected applications.