The average time it takes for a company to get a drug to market is fourteen years and the average cost is about $359 million, according to the February 1996 edition of Drug Discovery Today. Therefore, the speed with which a company passes through the various stages of the drug discovery process is critical to success. The pharmaceutical industry is constantly looking for new ways to reduce drug discovery time by identifying new targets and novel drugs to make the entire process more efficient and cost-effective. Combinatorial chemistry is one area in which tremendous advances are taking place.

One of the important stages of drug discovery is the screening, characterization, and validation of potential lead compounds. The process begins with lead generation and the design of huge libraries of compounds, whose chemical structures may interact with a target receptor and exert a desired biological effect. Library generation is followed by high-throughput screening (HTS) - receptor-binding assays that identify drug candidates or lead compounds by measuring their activity. When a compound successfully passes through HTS, it is considered a lead candidate and then progresses through characterization and validation studies known as lead optimization. Lead optimization is the fine-tuning of drug candidates for maximum effectiveness and minimum toxicity.

The roles of synthetic and computational chemists are interwoven in the process of lead generation and lead optimization. The expertise of synthetic chemists lies in their understanding of reaction and medicinal chemistry that helps them design and improve drug candidates. Computational chemists are experts in sophisticated computational algorithms and statistical techniques that help them build, subset, and focus huge libraries of potential lead compounds to produce better leads for synthetic and medicinal chemists.

The process of screening, characterization, and validation of new compounds is an area where technological advances have made a big impact. Several software companies offer solutions to facilitate the combinatorial chemistry process in areas such as library design and diversity analysis (Chemical Design, Daylight Chemical Information Systems, MDL Information Systems, Molecular Simulations, Tripos), robotics integration (Chemical Design, Daylight), provision of databases (MDL Information Systems, Oxford Molecular Group PLC), and handling and analysis of results from HTS (Daylight, MDL Information Systems, Oxford Molecular Group PLC).

THE COMPETITIVE EDGE
Combinatorial chemistry is a set of automated techniques used to synthesize large libraries of compounds and screen them for activity. The automation of this process saves companies valuable time by streamlining their synthetic programs, and it gives these companies a competitive edge by improving their capacity to generate novel drugs. All of this translates into huge cost savings and reduced time to market. Before combinatorial chemistry, synthetic chemists used to synthesize as many compounds as possible, with their performance at work often being judged purely on the number of molecules that they could make per week, month, or year. Combinatorial chemistry allows companies to make and test thousands of molecules in a fraction of the time.

Today, combinatorial chemistry is rapidly becoming a core technology for drug discovery in the pharmaceutical and biotechnology industries. However, although the benefits of these techniques have been expounded by many, combinatorial chemistry is still in its youth, and development continues. As acceptance broadens within the pharmaceutical and biotechnology industries, chemists are looking for ways to improve existing techniques. The emerging trend of companies like MSI and Tripos is to focus libraries more rationally on the biological target. This effort is being made as the research focus moves from quantity of compounds made to selectivity. "The focusing of libraries interactively for lead optimization is what many perceive as the next hurdle in exploiting combinatorial approaches for drug discovery," says Scott D. Kahn, director of Life Science Marketing at MSI. "MSI's combinatorial chemistry software uses both structure-activity information from analog data and structural information from protein crystallography and NMR spectroscopy to help focus libraries." MSI has been concentrating on determining structure-activity relationships (SARs) from HTS hits, using that SAR data in library design, developing new methodologies for focusing libraries, and validating these new methods.

FOCUSING A LIBRARY
How do you select representatives from a compound library that you can make and test in HTS? This problem is much like polling at election time. How do you choose the right people to ask, who fairly represent public opinion? In combinatorial chemistry, which compounds should you make in order to give good coverage of all the molecules in the library? The solution: Use diversity tools to take a representative sample of the most diverse compounds within a theoretical or virtual library.

Now, software is available that will select and focus extremely large corporate or commercial theoretical libraries, for example, software that combines one- , two- , and three-dimensional molecular properties to assess similarity and diversity. Diversity analysis introduces a rationale into the combinatorial chemistry process by using property data to design new libraries and reduce the final number of compounds that need to be synthesized. The ability to use shape and pharmacophoric activity descriptors facilitates the incorporation of SAR information into library design. The integration of SAR data allows the design of smaller focused libraries.

There is a sheer infinite number of descriptors that can be used to relate structure with activity, and most tools now provide comprehensive libraries of these. The search for new compounds, however, warrants the development of novel descriptors that capture the essential characteristics of the compounds. Tools to quickly develop such "customized" descriptors can provide a significant competitive advantage.

AN ITERATIVE PROCESS
Once you secure some positive HTS results, the task is to determine SARs using similarity tools, then use that information to focus on the most interesting compounds. You want to pick the best set of molecules that are as diverse as possible within the constraints of the SAR data. You can achieve this by using chemically expressive SARs provided by pharmacophores and shape descriptors and combining them with diversity tools. These tools calculate how well every molecule fits the common SAR pattern and scores the hits. You can discard the compounds you don't like, find the most diverse molecules within the remaining group, and repeat HTS screening.

This iterative process using SAR data is important in lead optimization and discovery of alternative drugs from existing treatments. It enables you to leap away from the confines of a single substructure lead compound, such as a common scaffold, and investigate more specific options. Using a steroid scaffold, for example, you would be limited to making and testing as many steroids as possible that may have only similar activities and may already be patented. In reality, the steroid itself may not even be important for activity; it may be the geometric relationship between a methyl group and a hydroxyl group that is responsible for binding to the target receptor. Introducing SAR data into combinatorial chemistry allows you to identify such relationships and discover entirely different molecules that may be cheaper and have a better toxicity profile and bioavailability. Even more important, you may improve your chances of finding a novel drug that can be patented.

THE VALIDATION ISSUE
As with all new methods, a key question is whether the answers are valid and reliable. Validation examines real experimental data to show not only that the techniques identify the correct type of compounds, but that they improve the process. In one example, Kahn started with a pool of seventy-five compounds selected from the literature, for fourteen known therapeutic targets such that the "library" contained a 4-8 percent representation for each activity. Kahn then used SAR data and iterative library focusing tools to select a subset of molecules that increased the chance of finding active compounds from 8 to 42 percent in the worst case and, in the best case, from 8 to 83 percent, a tenfold improvement. As these results are extended in scope, the importance of diversity tools becomes clear and convincing - you can minimize redundancy and increase efficiency by targeting your libraries. It has been estimated by researchers in the field that diversity analysis can increase your chance of finding active compounds by up to one hundred times.

Computational chemistry is undergoing a metamorphosis to meet the current and future needs of the industry. As applications become more widespread and the technology better validated, its use is spreading beyond computational specialists to the laboratory bench. This expert technology is becoming a means of transforming data into knowledge that can be shared among all of the members of a drug discovery team.

PLUG-AND-PLAYABILITY
But sharing expert analyses throughout an organization must be done in a way that is consistent with the corporate paradigm. As a result, technology is becoming more desktop oriented. Increasingly, new technology must be able to plug-and-play with the Microsoft productivity tools and be "Web aware." In a recent Chemical & Engineering News (6 October 1997) article, new and forthcoming Web-based products from computational chemistry software development companies were discussed. Several of these products were displayed at the last ACS meeting in Las Vegas.

New Web-based software is designed to meet the needs of synthetic chemists who would like to determine in advance whether a particular reaction chemistry would be a good one in terms of diversity analysis, and how best to combine it with other chemistries to make libraries. "Sophisticated modeling and analysis technology for computational chemists has existed for several years," says Kahn, "the task now is to move this technology into an environment where synthetic and medicinal chemists can use it." MSI sees that environment as the World Wide Web. WebLab products use the power of corporate intranets to increase access to proven simulation methods and to make them available to experimental scientists who tend not to be specialized molecular modelers.

The new software includes a series of tools that allows the synthetic chemist to assemble a library based simply upon reagents and the reactions that they undergo. The synthetic chemist can specify the reaction substituents and conditions and let the program automatically build the reaction. He or she can also specify the reaction chemistry for the second and subsequent steps in building a library. The synthetic chemist then may use validated expert diversity tools to select which compounds are worth making. The key difference with the expert computational tools available in combinatorial chemistry software is that it is Web-based and always expresses the molecules in terms of their reaction chemistry. This kind of desktop access to state-of-the-art validated technology gives synthetic chemists more autonomy, yet improves communication among members of the research group because information stored on a central server can be easily viewed and exchanged throughout the group.

This explosion of interest in Web technology has also resulted in the rapid development of proprietary Web tools that have been developed by pharmaceutical companies internally as part of corporate research programs. Introduction of these tools has forced the development of standard client-server communication protocols, thereby allowing the use of standardized computational engines and user-specific computational codes in intranet combinatorial chemistry applications.

MULTIDISCIPLINARY TEAMS
Several articles in Today's Chemist at Work (June 1997, December 1997) have detailed the limitless possibilities of the Internet and corporate intranets in facilitating better, faster, and more widespread communication and productivity. The accumulated validation of core computational chemistry technology over more than a decade has allowed the methods appropriate for specific applications to be preselected, preprogrammed, and precalibrated so that they can be used much like laboratory instruments. Thus, scientists who are not computational specialists can confidently use these instruments. This ability of Web-based applications to facilitate communication between multidisciplinary research teams is also demonstrated by software that is a suite of modules allowing medicinal chemists access to lead optimization tools that were previously only available to expert computational chemists.

A VISION FOR THE FUTURE
There is no doubt that interest in combinatorial chemistry remains strong. Not only has it been established as an effective paradigm for drug discovery, but it is evolving to address new issues and new applications, and this evolution is being matched by the development of new tools and methods.

It is clear that Web-based technology is the way of the future as a method to improve collaboration and communication within a multidisciplinary research team; you can cut and paste data and send structures easily to other team members. Intuitive Web-based applications bring a common language to a multidisciplinary research effort and an entirely new face to drug discovery for the pharmaceutical and biotechnology industry. The technological advances in the chemistry arena are no longer reaching only the expert computational chemists. Computational chemistry technology is broadening its reach to experts from other disciplines, thereby allowing them more freedom and greater opportunity to investigate projects more deeply.

SEE ARTICLE: Reshaping the Drug Discovery Environment