By Kevin Davies
May 12, 2005 | A sweetly aromatic odor hangs in the air over Tripos’ new chemical research center. The source, however, is not, as one might expect, the complex chemical syntheses inside, but a rather different form of organic rumination — as flocks of Jacob sheep safely graze nearby on England’s green and pleasant lands.
Bude (rhymes with ‘nude’), a quaint Cornish seaside town, is the unlikely setting for a 21st century chemistry laboratory charged with producing the next generation of small-molecule compounds for drug discovery, integrated with a raft of homegrown software solutions. The region is better known for pasties and surf — nearby Newquay hosted the world surf championships in 2003.
Such remoteness has its drawbacks — busy pharmaceutical executives won’t typically trek six hours from London unless they seriously mean business. But this bucolic setting has an upside too. Cornwall has given rise to many successful chemistry companies, and the easy-going, outdoor lifestyle definitely appeals to many of the staff.
Eager to move into chemistry, Tripos acquired the Bude facility in 1997, when it bought a small company with 12 employees called Receptor Research (see Sidebar ‘Interesting Times’). The site quickly expanded to more than 100 staff. After landing a four-year, $100-million deal with Pfizer to design and synthesize ultra-pure small-molecule libraries, Tripos further expanded the facility, resulting in the new Tripos Discovery Research Centre (TDRC), which occupies some 75,000 square feet and finds 130 chemists and dozens of programmers operating almost side by side. Last summer, His Royal Highness The Duke of York officially opened the new facility, dutifully donning a lab coat and touring the pristine labs.
During one of his frequent visits to Bude from his headquarters in St. Louis, Mo., president and CEO John P. McAlister justified Tripos’ push into chemical synthesis: “Many companies are saying that a large percentage of their [chemical] libraries are obsolete. They’re looking to find ways to first identify which ones are and which ones aren’t, and then replace the ones that are. I think these libraries need to be refreshed or replenished in an ongoing process.”
Pfizer’s more than $90-million ‘file enrichment’ deal with Tripos was designed to improve the size, quality, and diversity of its compound libraries. Pfizer has partnered with several other companies to enrich and diversify its chemical libraries, including ArQule, Discovery Partners International, and ChemBridge (see Sidebar ‘The Right Chemistry’).
At Tripos Discovery Research Ltd. (TDR), chemists routinely synthesize 8,000 to 10,000 compounds per week, but the “differentiator,” according to TDR’s managing director and senior vice president Mark Allen, is the nature of the compounds that can be made, their design, quality control, and ability to follow up on compounds synthesized. TDR has analytical capacity for up to 15,000 molecules per week. About 6,000 to 8,000 molecules are typically purified each week using high-performance liquid chromatography instruments from Waters and Agilent. About 90 percent of production is “fee-for-service” work, currently for Pfizer and a dozen other companies. The remainder fuels Tripos’ internal discovery efforts.
In total, some 200,000 compounds and 25,000 reagents are currently stored at TDR, all easily retrievable using a Kardex system, and easily scalable should kilogram quantities be required. That’s just a fraction of the company’s virtual library, which runs to tens of millions of compounds, according to Allen.
|Tripos President/CEO |
“People looked to Pfizer when they made the big investment in 1999 [with ArQule],” says McAlister. “Other pharmas were sitting back, thinking, ‘What are these guys doing?’ Then when Pfizer enlarged their investment with Tripos and others in 2002, many companies were still sitting on the sidelines, although beginning to realize their own problems. The word is getting out that this has been a useful exercise for them.”
Pfizer’s approach isn’t for everyone. GlaxoSmithKline, for example, invested approximately $300 million in reinventing its compound library facilities at facilities including Tres Cantos, near Madrid, and Harlow, England. “The jury’s still out,” observes Peter Hecht, TDR’s former managing director, who now runs Oridis Biomed GmbH, a tissue genomics company in Austria. “Do you want to invest hundreds of millions in automation and people, or do you go with the Pfizer model, where you’re focused on delivering things? If you work externally, you can sometimes get results quicker.”
Chemical synthesis is one essential component in what Tripos executives like to call “knowledge-driven chemistry”: how to reduce attrition rates using a knowledge-driven approach from disease to drug. If drug discovery is a cross-country car journey, then compound synthesis can be considered the car and the engine, while the IT infrastructure and software provides the maps and navigational system.
Maximizing TDR’s high-throughput efficiency hinges not only on great chemistry but also on a premium knowledge-management system. As Trevor Heritage, former senior vice president of informatics who joined Elsevier MDL in April, explains: “All of the samples look alike. The chemists spend lots of time trying to access reliable supplies of reagents as part of designing libraries — lots of time doing the research and validating certain reactions, reagents, etc. We needed to capture that information.”
The result is the ChemCore chemical tracking system, designed to capture all the relevant experimental data by unobtrusively mirroring the chemists’ protocols. “What we established in Bude was a way to understand the scientists’ workflow, and be able to customize the software to mirror the workflow,” says Heritage. “Most scientists won’t go out of their workflow to capture data the organizations want — particularly for reactions that fail!”
Tripos recognizes that different segments of the drug market are seeking different things. While Pfizer and other big pharmas might want a complete overhaul of their compound libraries, medium-sized biopharmas may be more interested in packages involving Tripos’ growing portfolio of internal discovery projects — Tripos is expanding its internal research program, focusing in areas such as G-protein coupled receptors or kinases, and believes that several discovery projects may appeal to pharma. And for smaller biotechs, Tripos will target chemistry services on a project-by-project basis.
McAlister says that establishing the right projects is vital for his clients. Based on internal models, McAlister says his firm’s technology can yield savings of tens, maybe hundreds of millions of dollars, on the overall cost of drug discovery, “including not only efficiencies gained in real time but also the effectiveness in killing projects [early] that aren’t going to succeed, then getting to market with a longer patent life on the compound.”
A challenge for Tripos is how to integrate TDR with its bulging portfolio of informatics products. “We didn’t believe that the tools we were producing were being used effectively,” admits McAlister. “The tools weren’t integrated into the process — people using the tools were separate from the people doing the chemistry or the biology. In fact, the way the organization was set up, there wasn’t a good way or incentive for them to interact effectively.
“We felt that if we were able to engineer a close interaction between the informatics and the chemistry people,” McAlister continues, “both from having the chemistry read out more effectively, and the informatics developed in a lab environment to make it relevant to the discovery process, that would help us market software more efficiently.”
For example, the ChemSpace tool (with lead-hopping activity) was developed in response to library design issues. “If you can make 500,000 or 200,000 molecules a year, which 200,000 do you make?” McAlister asks rhetorically. “You need a tool to help decide what’s the relevant parameter you should be looking for.
“By the same token, if you have a particular lead, then you want to build a library of molecules that are similar enough to have the same activity, but different enough that they’re not exploring the same space,” he says. “That’s a technology that we would not have developed had we not been doing chemistry at the same time.”
McAlister says having the chemistry in-house allowed Tripos to enhance the tool “to capture information about what I should make, but also: ‘Can I make it? How quickly can I make it? Do I have materials on hand to make it? Have I tried before?’ All these are questions most pharma companies don’t capture! They’ll do some design, they’ll say, ‘I’ll make this,’ but they don’t have a clue if someone tried to make it one year ago or six weeks ago, and if they made it, did it test positive in some assay?”
With the major investment at Bude complete, the focus is shifting back to informatics. A striking example is Schering AG, which has enjoyed a “highly efficient, trustful collaboration” with Tripos over several years, according to the pharma’s vice president, Rainer Metternich. When Schering executives saw the ChemCore system in action a few years ago while on an informational visit to Bude, they contracted to implement a similar system, designed to capture and disseminate knowledge about the experimental processes.
The result is ECIMS (Enhanced Chemistry Information Management System), which provides a globally used registration tool for Schering’s compounds, as well as the Tripos Electronic Notebook (TEN, scheduled for release as a standalone product later this year), and a tool for compound handling.
“ECIMS enables us to deal with registration of new compounds on a global level, company-wide — whether synthesized in the U.S., Europe or Japan,” Metternich explains. Registration information includes the reaction database “so the chemist can search for detailed information on specific reaction conditions. We make the know-how available to the whole organization.”
Schering employs a suite of Tripos programs in structure-based drug design, applied to target assessment, lead discovery, and optimization. One example comes in an oncology program — an emerging therapeutic area at Schering — screening against kinases. “We were in a position to optimize these leads in a very short period, with the help of these [Tripos] molecular modeling tools,” says Metternich. The resulting lead compound is already in Phase I trials.
Pfizer was a software customer of Tripos years before its file enrichment deal. The companies collaborated to build the Unity 3D chemistry searching system, as well as new informatics algorithms to provide new data analysis and pharmaphore recognition technologies. In addition to the file enrichment deal in 2002, Pfizer has funded development of software and algorithms to analyze high-throughput screening data, as well as visualization technology and structure description tools. These have been bundled into software packages called SARNavigator and Lithium, which were released last year. Adoptees include Aventis (genomics analysis) and TDRC itself, in some of its lead activities.
Despite anecdotal evidence that molecular modeling is having an impact, Heritage acknowledges it hasn’t had the impact it was hyped to have. “When I visit computational groups, they’re much more aware of the need to support medicinal chemistry teams than they were five years ago. Lithium helps molecular modelers address their biggest challenge — to improve the value of computational chemistry techniques,” says Heritage.
Lithium was developed as a communication collaboration tool between computational chemists and medicinal chemists. “One of the problems that pharmas have is that computational chemists are responsible for developing alignment models or QSAR [quantitative structure-activity relationship] models or protein models, but there isn’t a strong mechanism to publish those models to the rest of the organization.”
Tripos learned about this barrier facing medicinal chemists’ usage of computational chemistry from Pfizer. “They want to use techniques there and then — they don’t want to wander over to someone else’s lab,” says Heritage. Tripos hopes Lithium’s unique asset will be to help organizations gain greater value from investment in computational chemistry and molecular modeling techniques.
Released late last year, Lithium is being used by AstraZeneca, Merck (Germany), and several smaller biotechs. Of note are Lithium’s visualization environment and an array of capabilities, including a 3D structure editor that allows medicinal chemists to try their hand at molecular modeling. For example, a chemist can examine a receptor-ligand complex downloaded from an Oracle database, and explore alternative structures in an intuitive fashion, with the software handling the bond lengths, angles, and conformations. If a chemist does something weird, a traffic light signal turns red, alerting the user to rethink or seek additional help.
“Medicinal chemists are the ones who have to make the compounds,” says Heritage. “Having medicinal chemistry so far separated from molecular modeling itself has hindered those kinds of discussions.” Tripos is banking that Lithium will bridge that gap.
The latest acquisition to Tripos’ suite of software is provided by Optive Research, which was co-founded in 2002 by Robert Pearlman, chair of pharmacy at the University of Texas, and Bryan Koontz, who was also the company’s CEO. Tripos has marketed a handful of software products from Pearlman’s group over the past two decades. Following the launch of Optive, bringing the companies together — finalized in January 2005 — evidently made a lot of sense. It also made sense for Koontz to take over as senior vice president and general manager of discovery informatics, following Heritage’s departure.
Pearlman is best known for developing the Concord program — the industry-standard method for converting 2D chemical structures to 3D. Recent enhancements include more-accurate representations of the natural form of chemical structures, which is critical to accurately predicting a molecule’s docking properties. Another Optive program is EA Inventor, which allows users to design new viable chemical structures that meet a series of specified criteria. Optive also has desktop tools running under Windows to enable chemists to design and build libraries for themselves, which Koontz points out is very complementary to Lithium.
Another benefit of the merger was Optive’s established productive partnerships with Microsoft and HP, which Tripos hopes to build upon. “We’re working closely towards a benchware vision,” says Koontz, which is aimed at the lab scientist, enabling chemists to design their own libraries. “Imagine that!” Koontz exclaims. “[Microsoft] views the lab scientist as a highly specialized knowledge worker.... It’s a perfect match. It’s not a computer scientist who sits in front of the computer all the time.”
Koontz says the principal areas of collaboration will be in Web services and high-performance computing. Despite the dominance of Linux and SGI, “Microsoft wants some of that action,” says Koontz. “We benchmarked some of their programs and, surprisingly, we saw pretty incredible results from Windows, faster than all the flavors of Linux.” Tripos could help bring supercomputing to the masses, says Koontz.
It is no easy task integrating chemical synthesis and software. “It’s easy to set up high-quality chemistry, but how valuable is that in making the right compounds?” says TDR’s Allen. In the end, he says, it’s all about “bringing stuff to the chemist on a day-to-day basis. We’re developing programs to transfer knowledge from the expert to the fingertips of the practitioner. We’re aiming to bring knowledge to the chemist, to use that information to make the right compounds and the right decisions, faster.”