MOUNTAIN VIEW, Calif., March 25 - A crowd of 400 software entrepreneurs slumped into a lecture hall here this month to learn a few secrets of biology and drug discovery.
What they found was that unlike the traditional IT gigs they're more accustomed to, the land where informatics intersects with life sciences does not necessarily have a universal language.
Sitting in a small auditorium on the Microsoft campus, the crowd, drawn by the Silicon Valley Association of Start Up Entrepreneurs, stirred when panelists stressed the need for better software that could manage growing genomic, biological, and pharmacokinetic data.
One panelist, SciMagix CEO Robert Dunkle, who previously worked in chemoinformatic and bioinformatic businesses, fueled participants' interest by noting that hardly any drug companies have the tools to integrate their biological and pharmaceutical data.
By the time Fluidigm CEO Gajus Worthington, a veteran of the semiconductor industry, said he had raised $50 million in the last 3 years, one could almost hear the keystrokes of new business plans being drafted. Almost.
Dot-com-type entrepreneurs are used to a 2- to 3-year product cycle: Build it, market it, and move on to the next generation. Creating an informatics platform for drug development is not that different, the panelists promised. For one thing, the cycles are just longer and the platforms cannot be constrained. For another thing, faster and bigger is not always better. Targeted, yet flexible, applications are needed.
"We have to improve our prediction of efficacy and adverse events at the front end," said David Mack of Eos Biotechnology. "We're stuck with vertical silos of [drug] data that don't talk to each other. Most of the commercial [software] just hasn't delivered."
For example, despite its location in the heart of Silicon Valley, Eos licensed just one application for its platform to identify and validate new genomics-based drug targets. And though the firm built the rest of the software platform internally, Mack insists he's eager to invest in any technology that can reduce the number of researchers on staff.
The image informatics tools from SciMagix, meanwhile, offer a new twist on the data-integration problem. CEO Dunkle asserts that most data generated in life-science research comes in the form of images rather than letters and numbers. And those images are woefully underused by researchers.
SciMagix gathered a series of 2D electrophoresis gel analyses from a major drug company compound that six chemists had researched for years. Scientists understand that the presence of constellations in the gels can point to toxicity or efficacy--a task that still requires an experienced scientist to detect patterns. Within a month of applying a SciMagix image-analysis tool, the drug company was able to determine the compound was too toxic and based on a false hypothesis. The project was scrapped.
Data generated by target researchers have to be shared by cohorts on the compound side, said Dunkle. Such sharing is not automatic because the tools are insufficient. Image informatics is one tool to aid the process.
Panelist Lev Leytes, a founder of Virtual Arrays, pointed out most drug-discovery efforts still work sequentially. Research typically starts by pursuing the most potent compound. Yet the most potent also may be the most toxic, a problem that may not be discovered until a company spends millions of dollars verifying potency.
Virtual Arrays' new cell assays are designed to allow drug researchers to simultaneously perform parallel analyses in target validation, high-throughput cellular screening, new chemical entity selectivity and specificity, and early toxicology. Parallel discoveries will weed out non-promising compounds earlier.
Virtual Arrays is probably a year away from a product for "regular customers," but Leytes said a Peninsula-based biopharmaceutical firm already has contracted with his company to build an internal prototype.
Why can't there be standard algorithms, as in programming for electrical engineering and physics, which software engineers can use to write the proverbial killer app in drug discovery, asked an attendee? Isn't that the crux of the problem?
"Biotechnology is an inexact science," replied James Healy, managing director of Sofinnova Ventures.