Vertex: The Write StuffBy Kevin Fogarty | Posted 2004-11-01 Email Print
Re-Thinking HR: What Every CIO Needs to Know About Tomorrow's Workforce
A biotech company ditched its scientists' paper notebooks for a digital research system that speeds development of money-making drugs.
Mark Murcko's career is based on the idea that you can indeedplatitudes asiderush genius. In fact, his company's whole strategy is based on it.
So was his decision to take paper notebooks off the lab benches of his scientists.
As chief technology officer at Vertex Pharmaceuticals in Cambridge, Mass., Murcko's challenge is to accelerate the company's ability to produce a profit-making drug. The company has two lucrative HIV products on the market but is still bleeding cash after 15 years in the business. In 2003, it lost $196.8 million on revenue of just $69.1 million.
Murcko can't do a lot to accelerate the many rounds of clinical tests a new drug has to go through to be approved. But he can shorten the time it takes to come up with the drug in the first place.
Which is where doing away with notebooks comes in.
New drugs come from the benches of scientists who have to complete a bewildering list of tasks to develop even one good candidate. And they have to maintain records of every one of those steps to create a paper trail that adds legitimacy to a patent application, says Ian Smith, Vertex' senior vice president and chief financial officer. Without documentation, a patent application is worthless; without a patent, a promisingly lucrative proprietary compound could become public property.
Those records, until 2002, were kept in paper notebooks that the company's 100 chemists filled out by hand. They would note the types of molecules they were trying to synthesize, the effect they were trying to create and whatever progress they'd made.
Then they went electronicsort of. They would open an online product catalog to order the supplies they'd need to create the molecules, launch another piece of software to define the weight of the supplies, use another application to input the orders to machines that use robotics to automate the mixing of new compounds, and use yet another to create a visual image of what the molecule should look like. When all that was done, the chemist would compare, by hand and eye, the synthetic molecule with an image of the desired result.
If the molecule turned out to have the right properties, such as the ability to be absorbed easily by humans, the chemist would register it with the corporate database and store all of the data in a "supplementary materials file folder"an accordion file filled with printouts from a series of highly advanced informatics systems.
"So they were stuffing these accordion files until they filled one up; then they'd go get another one," says Steve Schmidt, the company's vice president of information systems.
The process took a lot of time.
Electronic versions of lab notebooks were available from suppliers such as EKM Corp. and CambridgeSoft, but most tried to re-create the same functions and drawbacks as paper notebooks, according to Schmidt. Others were too generic in their functions to do much good.
"We found that [electronic lab notebooks] need to be specific to the function you're supporting," Schmidt explains. "Chemists, biologists and [molecular] formulators use different databases and lab equipment. One size does not fit all.''
Vertex did time-and-motion studies on its chemists to find out how long it took to spec the supplies needed to create new molecules, give instructions to the robotic synthesizers and verify the results. The greatest amount of time was eaten up in switching among different systems to place orders and control the lab equipment, Schmidt says.
So when Vertex built its own electronic notebook, it tried to streamline the process.
The company's notebooks still let scientists keep their notes, albeit online and stored in an Oracle database. But the digital system also provides instant access to the commercial catalogs chemists use to buy the compounds they want to modify for a specific test. For instance, the chemists can search the catalogs for compounds that most closely resemble the molecules they're trying to create, and then test whether those compounds fit into a gene at a spot and in a way that achieves a desired result. By tracking this trial-and-error testing, the chemist creates a "library" of molecules from which to work. The researcher then picks the compounds that look most appropriate and runs them through another piece of analytic software to verify that the attributes are correct.
The researcher can then use a process-control application to create a set of orders for the robotic synthesizers, which actually mix and modify the compounds to match the patterns the researcher wants. The synthesizer, following yet another set of instructions, can then test all of the resulting compounds and compare them to a spectrogram showing what each should have looked like.
"So the chemist, instead of having to figure out what the compound is supposed to look like and see if it came out right, has a lot of help to verify that," Schmidt says. "I don't know why we call them notebooks. Internally, they're really [applications that track] process flows."
The end goal is to increase the flow of marketable drugs. And the system does this by increasing the work outputthe number of drugs that chemists can synthesize and testin a given year.
"We estimated that the help on execution was going to give us an improvement from 33% to 200% in productivity; we were very happy to find a fivefold increase in productivity instead," Schmidt explains. Instead of 75 potential drugs per month, each scientist can now test about 350.
In particular, the new software has helped accelerate the creation of two or three promising new compounds that are being submitted for patents, according to Schmidt. Details of what they are and what they're supposed to do have to remain secret until public testing makes disclosure unavoidable, he says.
Time Is Money
Such speed is critical if Murcko and the electronic notebooks are to turn Vertex' bottom line to black anytime soon. One billion-dollar drug would wipe out the company's financial woes.
Only about one in 5,000 new compounds turns into any kind of commercially sold product. And only a fraction of those that make it to market succeed on the billion-dollar scale. Even a modest drug takes 10 to 12 years to bring to market and costs somewhere between $240 million and $800 million.
That's why Murcko is obsessed with time. Before doing away with paper notebooks became a priority, he was instrumental in developing a "fast prototype" model that helps programmers put together new applications that can help chemists, in as few as three days. Once built, the prototype can be beat on (i.e., tested) by the bench chemists who would end up using it.
By fixing problems and tuning software on the fly, Murcko says, the chemists get tools tailored to their specific needs without waiting so long that the tool would become obsolete.
Among other tools developed using this method was the first version of the Vertex Research Database Interface (VERDI), which has for three years collated and presented relevant data from scientific journals to researchers as they need itsaving hours per day they would have spent searching for previous work on a particular molecule.
Vertex has been especially skilled at shortening its product-development cycle, says Joseph Pantginis, senior biotechnology analyst with bond broker J.B. Hanauer & Co.
"They have some expertise in the initial steps of the process in identifying the structure of the molecules; a lot of companies would have to outsource that," Pantginis says. "Once you do that , you can develop molecule A to fit in slot B."
Just increasing the number of compounds Vertex develops and tests doesn't help that much; the company also has to investigate the right ones.
The VERDI tool, which Schmidt describes as "basically just a knowledge management app," is credited with helping to do that.
"As you input data on a molecule you're synthesizing, information could pop up that will say that in 1983 Eli Lilly published a report in a journal describing that molecule, and would you like to see the report?" Murcko explains.
"Maybe a similar molecule hits a different receptorthat would indicate you should look for similar side effects," he adds. "We save time running down blind alleys that way; it keeps the research focused on the path most likely to be fruitful."
But in the end, even if it's something as simple as a notebook, Vertex often has to invent each and every wheel it rolls on.
"We see more and more widgets out there, pieces of software that each drug company has to string together to get the system to do what needs to be done," Murcko says.
"It's a young industry, and there's just not that much [commercial genomic research software] out there," says Craig Venter, who as CEO of Celera Genomics led a furious race with nonprofit researchers to map the human genome, a process that finished in a cooperative tie in 2000.
A Different Kind of Bottom Line
The need for Vertex to develop lab notebooks internally, even though commercial versions were shipping, is a prime example of that lack of software. And justifying the cost of development required a little creativity and faith, says CFO Smith.
"Over a short term [technology investments] are very difficult to justify," Smith says. "For a biotech company at the development stage, cash preservation is a key metric. ''
To justify an effort that might cost $200,000 or more, as did the electronic notebooks, Smith monitors items that don't figure in most CFOs' books, like the number of patents Vertex files, how quickly they're filed and how defensible they are.
"The first two patents we had challenged, on HIV products, we lost; now we're paying royalties on those," Smith says. "Recently we've had a 50-fold increase in patent filings, at greater speed, and they're more defensible [because they're documented more completely]; the VERDI system is largely responsible for that."
"We also count the time saved in building chemical libraries, the percentage of molecules we synthesize with the right properties to become drugs, and whether we save time for the chemistshow many hours a week of literature research they don't have to do," Smith adds.
The tools save about 5% of a chemist's total work time, giving the company the equivalent of five additional chemists and savings of about $1.25 million per year, Murcko estimates.
Vertex has two HIV drugs on the market, and another on the way. It also has several cancer drugs in development in partnerships with Merck and with fellow biotech pioneer Novartis. Two drugs that fight liver disease should hit the market by 2008. Vertex may be losing money, but it is sitting on around $500 million in investment capital. And it is shortening the time it takes to get a drug to market, Pantginis says.
"I.T. has really shortened the research process," he points out. "What used to take five or six years might take six months now. Vertex has a lot of expertise in identifying the structure of the molecules they're after, and once you do that, you're off to a really good start."
Vertex Base Case
Headquarters: 130 Waverly St., Cambridge, MA 02139
Phone: (617) 444-6100
Business: Pharmaceutical research and development
Chief Technology Officer: Mark Murcko
Vice President, Information Technology: Steve Schmidt
Financials in 2003: Revenue of $69.1 million; net loss of $196.8 million; research spending, $199.64 million.
Challenge: Shorten time to develop new drugs by improving productivity of bench chemists.
Vertex Pharmaceuticals: Working Smarter
Good drugs bring great profits to the company that develops them first. But drug development is slow and labor-intensive. Vertex Pharmaceuticals is saving time and labor by using a homegrown electronic laboratory-notebook application to automate the efforts of chemists who record every idea and experiment to document their discoveries. Another application saves time by culling research databases for relevant information, keeping scientists from having to repeat experiments others have already done.
Knowledge management increases quality of discovery. VERDI helped chemists disqualify unlikely compounds before wasting too much time on them.