SCIENCE/TECHNOLOGY
Computer-Aided Chemistry Edges Further Into Chemical Mainstream • Experimental chemistry and research needs are influences on software developments introduced at ACS meeting exposition i
208 ACS National
Meeting
Washington, D.C James H. Krieger, C&EN Washington evelopments in computer-aided chemistry continue to propagate and something of an electronic chemical consciousness appears to be taking shape. More and more, the disparate elements making up this evolving domain of chemistry are coming together in ways that present to its users an increasingly seamless web. This connectivity is moving computer-aided chemistry farther and farther away from the state of development where users had to adapt more or less rigidly to the confines of a particular software. Today, users are able to traverse the web with a fair amount of ease, determining for themselves those areas where the research department, the multidisciplinary project team, or the individual researcher wishes to reside and work. That's one image that emerged from the exposition held last month as part of the American Chemical Society's national meeting in Washington, D.C. The ACS national meetings and expositions have become a major forum for product introductions and announcements in computer-aided chemistry. Hence, this latest exposition provided a state-of-theart look at this rapidly developing area. What that look shows is a field characterized by:
D
ResearchStation for Windows provides an electronic workspace for gathenng, combining, and manipulating all available information about a project. • A modeling and computational chemistry discipline that is moving farther into the chemical mainstream as the needs of product research and experimental chemistry influence users and vendors alike. • Continuing efforts by software vendors to port their systems to more computer platforms so the systems can be run on Windows-based personal computers, Macintosh computers, or UNIX computers and workstations. • The increasing importance of open architectures for core software systems that enable users and third-party application developers to produce application modules for use with the systems. • The growing interconnectedness of laboratory research activities with molecular modeling, computational chemistry, chemical analysis, database searching, and general information retrieval. Many of these influences are evident, for example, in the new offering of Megalon. Focusing on scientific research in general, Megalon drew atten-
tion at the exposition with its introduction of ResearchStation for Windows, a scientific information management software package for personal computers. The company views ResearchStation as a core product that can run the vertical programs that Megalon has and will be publishing. Megalon is a relatively new company, formed by some of the people who had been involved with the molecular modeling venture launched by Autodesk Inc. a couple of years ago. Autodesk has been a major factor in computer-aided design through mass marketing of its AutoCAD system at relatively low prices. For its modeling venture, Autodesk had acquired exclusive sales and marketing rights to the HypeiChem molecular modeling system for personal computers with Windows from Hypercube Inc., the Canadian developer of that system. That association ended this past January, however, with return of the sales and marketing rights to Hypercube. SEPTEMBER 12,1994 C&EN
21
SCIENCE/TECHNOLOGY
Meanwhile, Megalon was formed as a Swiss firm based in Neuchâtel with a wholly owned U.S. arm in Novato, Calif. Manfred Maschek, chief executive officer of Megalon, likens the firm's operation to the functional behavior of a book publisher. It sees a need, seeks out a development group that has applicable software, puts the software through a quality assurance process, and transfers the technology into a product. The company's current products, for example, include ChemStructure for Windows (for publishing-quality output of chemical structures and images), Unistat 3.0 for Windows (comprehensive statistical analysis capabilities and presentation tools), and Compounds for Windows (that provides easy access to physical property and spectral data for 27,000 organic compounds). The company's marketing approach is to explore different sales avenues for more of a mass market. For example, it makes functioning product demonstration programs available on Internet via anonymous FTP (file transfer protocol). Maschek also says that the company's intention is to keep its products below $1,000 so that an individual scientist can afford them. In working out its product strategy, Megalon made an investment in Helix Systems, a Palo Alto, Calif., firm that was developing an information management software system. That project has now come to fruition in ResearchStation. Maschek views ResearchStation as the backbone for Megalon's entire output of vertical applications. 22
SEPTEMBER 12,1994 C&EN
ResearchStation employs Windowsstandard object linking and embedding OLE-2.0 technology to provide a dynamic electronic workspace for consolidating, manipulating, and examining all available information about a research project or topic. With it, scientists can gather, combine, and use information in any form, whether text, images, numerical data, graphics, or video and sound. The software serves both as a personal productivity and research tool and as a means of communication for collaborative work. Operating much like a laboratory log or notebook, ResearchStation provides a framework for collecting information from various sources—text from articles, structures from databases, raw data and images from lab experiments, analysis results and written documentation, and the like. In rearranging and examining the information, individual items can be viewed as icons or in outline form or can be opened to display contents. The items can be moved or altered with mouse commands, and text notes can be added for clarification or as reminders. In addition, ResearchStation provides a centralized environment for working with all desktop software applications. Minder is the name of a feature in ResearchStation that provides tools for automating routine and repeated tasks such as searching for information, creating and routing reports, updating files, and signing off on projects. Another feature of ResearchStation is that it creates an audit trail that allows information to be organized on the basis
Molecular Simulations (left) and CAChe Scientific were among many software vendors at the ACS exhibition. of content, history, or relationships to other information. For example, a data log for each document automatically tracks ownership, activities, and modifications to the document. And ResearchStation can supplement the automated data log and audit trail with specialized procedures for security and standards requirements. For example, electronic sign-off can be used for approval of completed tasks, locking work, and tracking the routing of files. On another front, Molecular Simulations Inc. (MSI) is an example of a molecular modeling and computational chemistry software firm that has been fleshing out its approach to the wider research market. At the ACS meeting in San Diego in March, the company introduced version 1.0 of a new environment and product line, Cerius2. It has since added to the line. Cerius2 has an architecture that incorporates MSI's chemistry backplane and an open, object-oriented software environment that other software modules can plug into easily. The heart of the system is C2· Visualizer, which provides the graphics visualization and modeling tools. By plugging particular software modules into Cerius2, users can tailor the system to particular research needs.
Version 1.0 of Cerius2 concentrated on modules for materials research—in catalysis and separations, crystal growth, and polymers. In July, the company introduced version 1.5, which brought life sciences functionality to the system with a module called C 2 D r u g Discovery Workbench. At the ACS exposition, the company introduced C2· Polymorph, a module that predicts polymorphs of molecular crystals. C 2 Drug Discovery Workbench and 2 C · Polymorph illustrate the approach that MSI has been developing for its offerings in computer-aided chemistry. According to MSI president and chief executive officer Michael J. Savage, the company is working on a strategy to expand the market by developing "solutionsoriented" products. He views the emerging strategy as leading to a "new generation of analytical instrumentation companies," where the instruments are virtual instruments rather than actual ones. MSI, Savage says, aims to address the range of needs across the continuum of the technology life cycle in computational chemistry. At the beginning of that cycle, he explains, research scientists, most often those at universities, are constantly discovering and developing new scientific methods, which at that stage have high scientific interest and high potential utility. The research scientists move on to new challenges, while the methods, with initial validation, are employed by computational chemists in industrial laboratories where they are further validated on commercially important problems. The methods are then in a position to be applied by a broader range of users to a broader range of products, thus increasing their practical utility and commercial value. Modeling software companies traditionally have focused on the area of the cycle with high scientific interest. But that area of the market became saturated, while increasing computer power and software that was easier to use were generating greater interest at the commercial value end of the cycle. To meet new market needs and possibilities, software firms—MSI among them—have moved beyond the "interesting science" phase to the "business-decision" phase. The three key levels to Cerius2, Savage explains, match the different areas of the life cycle continuum. At the most basic level is an open software developers kit (SDK), which can be used by academics in developing their applications. At the
next level are methods, such as those for molecular mechanics and dynamics simulations. At the highest level are the socalled computational instruments. Savage draws an analogy likening the latter to traditional analytical instruments: A traditional instrument takes a sample, measures a property, and outputs results; a computational instrument takes a molecular model, computes a property, and outputs results (solutions to a problem). C 2 Drug Discovery Workbench contains a variety of tools to quickly and automatically build molecules and congeneric series, perform conformational analyses, generate a large variety of two- and three-dimensional molecular descriptors, align sets of molecules, determine 2-D and 3-D quantitative structure-activity relationships (QSARs), and construct pseudoreceptor models. There are plans for future release of modules for database access, molecular field analysis, advanced alignment tools, and a variety of de novo ligand design methods.
C2 · Polymorph, Savage says, begins a move downstream. It provides a novel approach, he explains, to solving a problem costing industry millions of research dollars. Polymorphism is the ability of a compound to exist in more than one distinct form, and it affects numerous processes and product formulations, from pharmaceuticals and agrochemicals to pigments and dyes. Polymorphism can radically alter properties such as bioavailability, solubility, bulk handling, and shock sensitivity. C2· Polymorph—which is based on the Polymorph Predictor method developed by the research group of Ciba-Geigy scientist Heinrich Karfunkel in Basel, Switzerland—is designed to help companies select the optimal polymorph for product patenting, registration, and processing. The exposition also marked for MSI the first showing of combined products since the mid-July merger of BioCAD and MSI. BioCAD's basic product was Catalyst/Hypo, a hypothesis generation system for the company's Catalyst
A (claisen
SENSATION
When it comes to Claisen condensations, DBE-4 (dimethyl succinate) is a natural. Which is one reason why it'sfindingwide usage as a starting material for quinacridone pigments. It's also an important intermediate for insecticides, pharmaceuticals and specialty chemicals. In fact, the list of patents just keeps on growing. Could yours be next? There's only one way tofindout. Call 1-800-231-0998 for a sample.
DU PONT DuPont DBE Intermediates © 1993 DuPont Company
CIRCLE 4 ON READER SERVICE CARD
SEPTEMBER 12,1994 C&EN 23
SCIENCE/TECHNOLOGY drug discovery software. Catalyst/Hypo, Savage says, complements the C 2 Drug Discovery Workbench and allows MSI to provide a powerful set of drug discovery tools to medicinal chemists as well as to computational chemists. A companion product, Catalyst/Info, provides 1-D, 2-D, and 3-D chemical database searching capability. Change was also afoot this summer at Tripos—now Tripos Inc., following the St. Louis company's separation in June from parent Evans & Sutherland Computers Co. and its transformation into a publicly held company. At the exposition, Tripos announced that it is shipping a major new release of its Sybyl molecular modeling and analysis software on Silicon Graphics and Sun workstations. The new version 6.1 unites the family of tools for modeling, information management, and analytical processing under one central umbrella. "This release is an important milestone for us since it fulfills the vision of a unified research information management/ analysis system," says Matthew Clark, Tripos scientific market manager. "The central umbrella is key to establishing a unified research team." Besides the new version of Sybyl, Tripos released Capri, an option to its Triad nuclear magnetic resonance processing software. Applicable to any type of structure, from small natural and synthetic compounds to large peptides, proteins, and nucleic acids, Capri automates multidimensional NMR resonance assignments and accurately assigns atoms in the corresponding molecular structure. Tripos also announced the imminent release of the first tools for computeraided mass screening. These are aimed at researchers involved in combinatorial chemistry programs. According to Tripos, more than half of the world's pharmaceutical and biotechnology firms have initiated some type of combinatorial and mass screening effort. Combinatorial chemistry is the name given to synthetic strategies that produce libraries of thousands to millions of compounds to screen for promising drug leads (C&EN, Feb. 7, page 20). Mark W. Schwartz, vice president for marketing at Tripos, notes that mass screening for combinatorial chemistry involves two parts. The first is that of generating, storing, searching, and retrieving the myriad compounds. The second is doing something with the information once it is in hand. 24
SEPTEMBER 12,1994 C&EN
Tripos announced imminent release of software for combinatorial chemistry.
Tripos' approach to the first phase is well in hand, Schwartz says, and should be released in a few weeks. It will include the ability to "explode" a combinatorial query into all possible structures and store the information in libraries such as a Tripos Unity database or Molecular Spreadsheet. The libraries can then be screened for new leads. Tools for the second phase, applying diversity and similarity analysis to the libraries, are well along in development, Schwartz adds, and should be ready by the end of the year. Schwartz also notes that in another effort Tripos is working to apply the company's expertise in a number of areas to make it available for experimental chemists. The product would take advantage of a client-server approach and provide for chemical information management as well as chemical information analysis—involving analysis for patterns and trends, for example. From its beginnings, CAChe Scientific, Beaverton, Ore., has aimed its offerings at the experimental chemist. For example, its GroupServer cross-platform network system allows work group members using personal computers to share more powerful resources on servers such as IBM RISC System/6000 and Silicon Graphics workstations. And its ProjectLeader application is designed to simplify project planning, experiment setup, program execution, and documentation for the experimentalist. At the exposition in San Diego last spring, CAChe Scientific and IBM announced a partnership agreement under which the chemical services and applications unit of IBM's research division in
San Jose, Calif., would develop quantum chemistry applications that would utilize the CAChe system and be marketed by CAChe Scientific. That agreement has now borne fruit. In Washington, the parties introduced a new quantum chemistry program that will be available this autumn and that is designed especially for determination of molecular structure, properties, and reactivity of larger molecules. The program has since been named Mulliken, in honor of 1966 chemistry Nobel Laureate Robert S. Mulliken, widely regarded as the father of molecular orbital theory. "The software," says CAChe Scientific president George Fabel, "enables our customers to use more advanced techniques to generate even more accurate estimates of chemical structures, providing a better starting point for property estimates and calibration of experiments. Moreover, parameters for first-row transition metals make more accurate modeling of metal complexes a reality." IBM's A. Currie Munce, manager of scientific and technical application software at the company's Almaden, Calif., lab, notes that "practical chemical problems involving more than 100 atoms will now be solvable in standard industrial computing environments in less time than was ever before possible." The program provides four methods to study molecular properties, reactivity, and structure—namely, molecular mechanics, semiempirical quantum chemistry, ab initio quantum chemistry, and density functional theory. The program also incorporates what IBM calls ArChem, a new architecture for chemistry. It is an open, well-documented interface, allowing in-house or
third-party computational modules to be easily interfaced with ArChem. Speed comes from the program's ability to parallelize a problem. A user with workstation clusters or parallel processors can simply indicate for the software the names of the computers it should use in running the problem. The program automatically distributes the problem to them. An example used by IBM and CAChe Scientific to illustrate the program's capabilities involves valinomycin, an antibiotic molecule consisting of 171 atoms. The charge density of the molecule can be calculated on an IBM RS/6000 workstation in about 90 minutes. Running the calculation on three parallel computers would reduce the run time to under 40 minutes. Speed, the developers explain, is also derived by taking advantage of molecular symmetry. For example, the charge density of buckminsterfullerene can be calculated in only six minutes. Without symmetry, the calculation would take about eight times longer. CAChe Scientific also provides a current example of the continuing movement of computer-aided chemistry software to more and more platforms. At the exposition, the company announced a new generation of its WorkSystem software based on Apple Power Macintosh computers to take advantage of the high-performance PowerPC microprocessor. Originally introduced in a Macintosh version, the software was later brought out for Microsoft Windows, as well as in versions for Silicon Graphics and IBM RISC System/ 6000 workstations. "The PowerPC-based machines are an important development for molecular and reactivity modeling," according to CAChe Scientific vice president of engineering Richard DeHoff. "These powerful RISC systems provide a remarkable level of price and performance. In fact, the PowerPC performance is up to twice that of competitive RISC machines in our standard computational chemistry benchmarks." Another current example of expansion to new platforms comes from Hypercube. The company announced at the exposition that on Sept. 1 it would begin distribution of its HyperChem molecular modeling package in a Windows NT version for Alpha AXP systems from Digital Equipment Corp.
"Running on Digital's Alpha AXP PC, HyperChem for Windows NT provides our customers with five times the performance of a 486-based personal computer, and two- and one-half times that of a Pentium-based PC," Neil Ostlund, Hypercube president and chief executive officer, explains. The Windows NT version of HyperChem is release 4, the new offering announced by the company last spring following its withdrawal from the agreement with Autodesk. HyperChem had been the first chemistry application on Windows, Ostlund points out. But he thinks more can be done, and the company is now building UNIX products for platforms such as the Silicon Graphics Iris 6000. The company is also getting into networking and has created a license manager that allows it to provide schools with site licenses. Ostlund views HyperChem as a core around which he would like to see "an industry of applications" built. For example, he says, a crystal could be
created in a crystal builder product and sent by dynamic data exchange (DDE) into HyperChem. Hypercube has its own package of applications called ChemPlus that includes nine HyperChem extensions such as conformational searching and crystal building, and in June the company began distribution of HyperNMR, a software package for the a priori prediction of NMR spectra. HyperChem also now interacts with MDL Information Systems' ISISDraw and ISISBase directly, and according to Ostlund, ChemDraw from Cambridge Scientific is probably next on the list. Says Ostlund of Hypercube and HyperChem, "We're back in in spades." A new platform offering also figures in developments from Cambridge Scientific Computing, Cambridge, Mass. The company chose the exposition to launch its CS ChemDraw Pro for UNIX. A software program for drawing chemical structures, reaction schemes, and reaction mechanisms, CS ChemDraw Pro had been available in Macintosh
5-CARBON
Misfit
We admit that the chain length ofDBE-5 ^M (dimethyl glutarate) makes it a bit unusual. But that's precisely why it's so good for improving polymerflexibility.And for making plasticizers with superb low-temperature performance. Got any other odd thoughts? Call 1-800-231-0998 for your sample.
DU PONT DuPont DBE Intermediates © 1993 DuPont Company
CIRCLE 5 ON READER SERVICE CARD
SEPTEMBER 12,1994 C&EN 2 5
SCIENCE/TECHNOLOGY and Windows-based personal computer versions. Jonathan S. Brecher, CS direct marketing manager for Cambridge Scientific, notes that the company views the UNIX version as rounding out its ChemDraw family. It allows users to have tighter integration of ChemDraw with some of the higher end modeling programs, he points out. The exposition was also the venue chosen by Cambridge Scientific and Beilstein Information Systems to announce a new partnership aimed at developing and marketing solutions for the needs of the research chemist and the chemical information specialist. ChemDraw comes as part of an integrated suite of programs called CS ChemOffice that also includes Chem3D and ChemFinder. Beilstein Information Systems, based in Frankfurt and with U.S. offices in Englewood, Colo., publishes the "Beilstein Handbook of Organic Chemistry." It also offers the Beilstein Database, with information on more than 6 million organic compounds and associated properties, searchable with the firm's CrossFire high-performance search software. The first development to result from the new partnership is CS ChemOffice 3.0 Gateway for CrossFire. Scheduled to be available this fall, Gateway will provide access to the Beilstein Database by linking RISC data servers with clients that use Macintosh computers. Another recent development on the information front—though not connected with the Washington exposition—teams MDL Information Systems, San Leandro, Calif., with Dialog Information Services. The two firms announced an agreement last month to jointly develop software applications in the chemical area. The applications will use MDL's client-server ISIS software to access and search scientific and competitive information on Dialog together with a customer's proprietary information systems. In other developments at the ACS exposition, show goers had their first look at ChemNovel from the U.K.'s Chemical Design, which has a U.S. office in Mahwah, N.J. Developed as an integral part of the company's Chem-X modeling and computational software, ChemNovel is a tool for lead generation to help in the discovery process for new active drugs. 26
SEPTEMBER 12,1994 C&EN
Hardware advances lead to new graphics potential Although software advances dominated molecular modeling and computational chemistry exhibits at the ACS exposition, chemists there were able to see some results stemming from hardware advances as well. For example, Silicon Graphics Inc., Mountain View, Calif., sported a new microprocessor, the 64-bit RISC MIPS R8000 chip set, that when incorporated in the company's Power Challenge server can nearly bring the power of a supercomputer to the desktop. The number-crunching capability of the new chip can be combined with SGI graphics accelerators to generate graphics images that display very quickly onto the screen, boosting graphics capabilities to new levels. Mark Berger, chemistry market manager at SGI, notes that researchers in the company have been exploring this new capability to see what potential there may be for molecular modeling. One example comes from SGI molecular graphics expert Michael Teschner, a member of the technical staff in the company's Basel, Switzerland, office. He applied the concept of texture mapping to a model of the polypeptide antibiotic gramicidine A. Working with Teschner were Christen Henn of the Maurice E. Mueller Institute for Microscopy at the University of Basel, Juergen Brickmann of the University of Darmstadt in Ger-
ChemNovel uses descriptions of the pharmacophores of a compound— those structural features required for particular activity—to invent new molecules that meet the requirements for such activity. Using a scoring strategy, ChemNovel rejects those molecules
many, and Peter Flueckiger of the Visualization Group CSCS in Manno, Switzerland. The texture mapping work has been published in the Journal of Molecular Graphics [12, 98 (1994)]. Use of texture mapping for color coding surface properties allows the modeler to combine multiple channels of information into one display. For gramicidine A, the technique leads to a combined display of the electrostatic potential (ESP), indicating how specific parts of the molecule may interact with other molecules, and the molecular lipophilic potential (MLP), which gives an estimate of where the molecule has contact with either water or the membrane. One image (left) shows the application of a two dimensional texture map to visualize ESP and MSP simultaneously. The surface is color coded against ESP, with MLP acting as a filter to ESP so that the coding is active only at those locations where MLP surface values are above a certain threshold. In the other image, a computed property replaces information filtering to clip away the surface at the property values of the texture which should act as a filter. Teschner stresses that the clipping is done during the rendering step through the texture-mapping mechanism. The original surface data are left unchanged.
that are impossible to synthesize and down-weights those containing unfavorable components. Keith Davies, technical director of Chemical Design, notes that the scoring of molecules is based on how synthetically feasible they are. Various features,
such as flexibility and joining of fragments, are used to compile the score. Davies explains that the scoring is abstracted from the "Dictionary of Drugs" published by England's Chapman & Hall so that there is a fair assurance that the software program is generating druglike molecules. Polymers are the target of a new effort at Biosym Technologies, San Diego. The company is adding a new consortium that will augment others in the company's development strategy for its molecular modeling and computational chemistry software. Called the Block Copolymer Project, the consortium is intended to address the needs of industry for programs that can prediet the structure and properties of thermoplastic materials. As a long-term strategy to help guide its development efforts, Biosym has formed consortia with members from drug, petroleum, and chemical companies; government laboratories; and universities, working with dedicated teams of Biosym scientists and programmers. Each consortium focuses on a specific area of industrial technology. Member companies participate in design decisions to make sure that resulting products are relevant to their commercial problems, meet regularly, receive quarterly reports, and have access to Biosym scientists for consultation. Currently, Biosym has four consortia—for potential energy functions; polymer; catalysis and sorption; and electronic, optical, and magnetic materials—involving more than 100 chemical and pharmaceutical companies. The Block Copolymer Project will focus on morphology prediction, composite theory, and interactive graphics to create software that will provide new insights into the structure and properties of block copolymers. Companies that participate in the project, Biosym says, likely will have an interest in coatings, solvent or hot-melt adhesives, thermoplastic elastomers, urethanes, blend compatibilizers, and rubber-toughened thermoplastics. The consortium is planned as a twoyear project. As is Biosym's practice with other consortia, the members will receive the new programs as they are developed, ahead of commercial availability. A number of new developments were announced at the Washington ex-
position by SoftShell International, Grand Junction, Colo. The company markets the chemical drawing software Chemlntosh (for the Macintosh) and ChemWindow (for Windows). New network versions of ChemWindow and Chemlntosh are now available. Two of the licenses feature an authorized-access policy, while a third is designed for student laboratory networks. Also available is a new tool, MS Calculator, that enables users to calculate all possible formulas for a given mass simply by entering the mass and clicking. Constraints on element counts, mass percent for an element, and the number of double bond equivalents can also be entered, and the list of possible formulas can be sorted or copied to the clipboard. Set by SoftShell for release in November is a new chemical database software package called Entropy. Entropy stores structures and text on entry pages that require no formatting, and users can import their old files from Chem Window, Chemlntosh, and
other chemistry drawing products to create an instant database. Users can search on text, structures, or substructures. SoftShell points out that Entropy requires ChemWindow or Chemlntosh and shares much of the same user interface. Judging from the Washington exposition, this summer has been a busy one for the computer-aided chemistry community. In users' hands, all of the new systems and software will undoubtedly be churning out a profusion of reports and presentations. The timing may thus be propitious for SoftShell's final new product released last month: a new clip art collection, the Art of Science, prepared especially for the scientific community. SoftShell notes that the images, in PostScript format, can be used with most page layout, presentation, and word-processing programs. Included are images of scientists and technicians, scientific equipment, animals, insects, planets, spacecraft, and scientifically oriented cartoons. •
r%. Polyester
rizmzL Ifyou 're looking for a highly reactive source of the C-6 moiety for condensation polymers, look no further than DBE-6 (dimethyl adipate). You'll enjoy low reaction temperatures. Minimal side reactions. And, because it's a non-hygroscopic liquid, easy handling. What's more, DBE-6 excels in Dieckmann condensations. Anything else? That's up to you. Call 1-800-231-0998 for your sample.
DU PONT DuPont DBE Intermediates © 1
