Those despairing that the biotech market has lost some of its luster in the last few years can take heart in the recent decision by telecommunications firm Direct Wireless to rebuild itself as a drug discovery firm. Compared to the embattled telecom sector, which appears to have gone south for good, drug discovery has offered the struggling firm a relative aura of optimism that guided its choice to ditch its original business plan and start over from scratch.
In an additional glimmer of hope for the struggling bioinformatics sector, computational biology is proving to be the key component in Direct Wireless’ marketplace makeover: The company first signaled its intention to morph from its original model with the acquisition of bioinformatics consultancy the Barnhill Group, and followed up a week later by acquiring bioinformatics software firm Fractal Genomics.
“This is just the beginning of the technology portfolio,” said Stephen Barnhill, founder of the Barnhill Group and president and medical director of the new company, which will relaunch on Sept. 15 under the name Discovery Health Corporation. “We’ll be looking for technology both in the analytical space and in the bioinformatics space,” he added.
Barnhill, a pioneer in the use of neural networks and support vector machines for life science research, holds a number of patents in artificial-intelligence-based machine learning techniques and diagnostic discovery (see box, this page). Prior to founding the Barnhill Group, he was founder and CEO of Barnhill Clinical Laboratories, which was eventually acquired by Quest Diagnostics.
Direct Wireless was originally founded to develop a new wireless communications technology, but according to a statement from CEO Bill Williams, “the negative events that have occurred in the communications industry have made it difficult for us to advance our communications platform.” The company needed an additional $10 million to commercialize the technology — an amount that proved unattainable for the over-the-counter publicly traded firm to raise in the telecom freefall.
With the goal of changing the company’s strategic direction to drug discovery, Williams opted to acquire the Barnhill Group’s assets and place Barnhill at the helm of the new entity.
“I’m really creating what was the Barnhill Group within Direct Wireless,” Barnhill told BioInform.
Working within the shell of an OTC firm gives the nascent company several advantages, Barnhill noted. “Because it’s a public company, there are ways of raising capital to do this kind of research and we’ll be pursuing some of those in the future,” he said.
Direct Wireless has raised $364,816 in net cash from financing activities since its inception in April 2001, according to the most recent documents filed with the SEC.
One of Barnhill’s first moves as head of the new firm was to acquire Fractal Genomics, a San Francisco-based software startup founded by Sandy Shaw, a former bioinformatics specialist at the University of California, San Francisco. The company’s Fractal Genomic Modeling software, which Shaw described as “a combination of clustering and modeling at the same time,” is able to model patterns of similarity in large biological databases using a mathematical method based on fractal geometry. The software can be used to reverse-engineer genetic networks using less computation than other methods, Shaw said, and is particularly useful for time-series data.
Last week, the company provided proof of concept for the technology at the International Society for Oncodevelopmental Biology and Medicine conference in Edinburgh, UK, where Herbert Fritsche, from MD Anderson Cancer Center, presented a set of seven genes he found using the software that he said can separate acute lymphocytic T-cell leukemia from acute lymphocytic B-cell leukemia with 100 percent accuracy. Fritsche found the genes during a test of the software on the well-known ALL/AML leukemia dataset from Golub et al. [Science Oct. 15 1999: 531-537].
Barnhill said that the company also used the Fractal Genomics software in a collaboration with Fritsche to build a “time series causal matrix” for the progression of lung cancer from early stage to late stage. “By doing time series we can see how genes are changing and affecting each other,” he said.
Researchers at St. Jude Children’s Hospital and the Alvin J. Siteman Cancer Center at Washington University are also using Fractal Genomics’ software, Shaw said.
Starting with Better Data
Despite its strong foundation in computational biology, Barnhill stressed that Health Discovery Corporation will not rely solely on bioinformatics. “We don’t just do the computation,” he said. “To me, the computational piece doesn’t start with data. If you’re going to really do this right, the computational piece starts with the problem, and our computational biologists and mathematicians will get involved right from the inception of the identification of the clinical dilemma that’s to be studied.”
The company plans to make experimental design a key part of its partnership strategy, advising its clients on “how many specimens they need for clinical significance and statistical significance, how many they need, what disease types they need to be, the range of the disease, the stage of the disease, to make that discovery for them,” Barnhill said.
This focus on data generation as well as data analysis should set the new firm apart from its computational competitors, Barnhill said. “Other computational companies are really taking the ball from the data, and it’s like the old computer adage: garbage in, garbage out.”
Barnhill said that the company is currently in negotiations with potential pharmaceutical clients.
Stephen Barnhill’s Patent History
- US Patent 5,769,074. Computer assisted methods for diagnosing diseases.
- US Patent 6,128,608. Enhancing knowledge discovery using multiple support vector machines.
- US Patent 6,157,921. Enhancing knowledge discovery using support vector machines in a distributed network environment.
- US Patent 6,248,063. Computer assisted methods for diagnosing diseases.
- US Patent 6,306,087. Computer assisted methods for diagnosing diseases.
- US Patent 6,427,141. Enhancing knowledge discovery using multiple support vector machines.
- US Patent Application 20030023571. Enhancing knowledge discovery using support vector machines in a distributed network environment.
Sandy Shaw’s Patent History
- US Patent Application 20010047376. Method for the manipulation, storage, modeling, visualization, and quantification of datasets