The care and feeding of geniuses is something Ive been doing for 30 years, says Peter Farley, CEO of Savannah, Ga.-based Biowulf. Case in point: as CEO of the original biotech startup, Cetus, Farley greenlighted Kary Mullis idea for PCR in 1985. These days, the industry pioneer is betting that the next revolution in biotechnology is going to be mathematics.
Using Biowulf as a virtual enterprise, Farley has assembled a corps of theoretical mathematicians who focus on machine learning computational statistics-based math using N-dimensional geometries to parse the features of large datasets.
Momentum for the company (which was developed in 100 percent stealth conditions, says Farley) picked up when the CEO successfully recruited Vladimir Vapnik, whom he calls the Einstein of machine learning. Farley then asked Vapnik to list other mathematicians who would advance the field. Of the 24 researchers he named, 18 have signed on to Biowulf, some leaving posts at Microsoft and Bell Labs. (One of the people named was David Haussler, who did not join Biowulf but did lend his support in securing Vapniks visa to come to the US.)
Farley contends that his so-called Math-topia will be the next biotech frontier. As proof, he presents unpublished results that identified specific cancer-promoting genes (based on two commonly used data- sets) more precisely and conclusively than other paradigms such as neural nets or hidden Markov models.
The business model for Biowulf is fee for service. The company hopes to establish itself as the global gold standard for data mining, and in doing so to reap monopoly premiums from clients. One possible Achilles heel: for Biowulf to succeed at lucrative problems in fields like proteomics, its facilitators will have to bridge the clients to the mathematicians, a group not known for collaborative wherewithal.
Still, Bernhard Schoelkopf, one of Vapniks protégés and now a member of the Biowulf braintrust, is excited. I think we probably have the best machine-learning and [support vector machines] group in the world, he says. For a startup, thats pretty amazing.
Brad Stenger