Chemical & Engineering News

May 12, 1997


Copyright © 1997 by the American Chemical Society

COMPUTATIONAL CHEMISTRY IMPACT

New developments in underlying software propel technology into activities spanning discovery cycle

James H. Krieger
C&EN Washington


From the ACS meeting

Computational chemistry is growing up. No longer simply a rambunctious technology with promise, it is moving into the mainstream of the chemical enterprise.

Core computational chemistry and information management software technologies are now solidly in place. The Internet's World Wide Web technology is providing a new and growing means of disseminating software tools, both through the Internet and through corporate and organizational networks. And an extension of computational chemistry's purview to bioinformatics - currently a hot area - is now spurring the development of software for the front end of the drug discovery R&D sequence covered by computational chemistry. At the other end and in between, new application software is further helping to provide tools that span the discovery cycle.

Meanwhile, a great deal of software development activity is being expended on the software firms' core computational chemistry programs. Indeed, new versions of many flagship software packages have just been released, and they and some entirely new programs are bringing new levels of capability, flexibility, speed, and accuracy to the application of the technology.

On the business side, alliances in one form or another - both between software development companies and between those companies and their chemical and pharmaceutical industry customers - continue to grow. Such collaborations increasingly characterize the software development business. And what might be called consulting activities by the software firms have strongly taken root and are becoming more than an appendage to the software development part of the business, especially among the larger software companies.

Michael J. Savage is president and chief executive officer of computational chemistry software developer Molecular Simulations Inc. (MSI), San Diego. As he puts it: "Computational chemistry is really becoming a tool that's a must-have tool rather than a nice-to-have tool."

All of these developments and more were in evidence recently in San Francisco at the exposition held in conjunction with the national meeting of the American Chemical Society. Over the years, the ACS expositions have become the predominant showcase for computational chemistry software.

It's a field that over those same years has come to encompass an array of activities that include computational chemistry, molecular modeling and simulation, and chemical information delivery and management. As often as not, such activities are interrelated.

For the past year or so, World Wide Web technology has held an increasingly prominent place in the development activities of the software companies. Besides providing the familiar Internet entre to company home pages, the technology represents a new way to handily disseminate computational chemistry software tools to widespread groups of users.

The technology underlying the web enables users to access properly coded documents or programs on server computers by employing browser software. Browsers such as Netscape Navigator from Netscape Communications, Mountain View, Calif., and Internet Explorer from Microsoft, Seattle, are popular examples.

None of this browser-based technology is limited to the World Wide Web, however. The same technology can be deployed on an organization's internal network, turning it into a so-called intranet that can serve up documents and information stored on the network's servers. And because web technology works with all kinds of computer hardware, organizations can network their often disparate collections of computers and make information readily available within the organization. Indeed, that platform independence has helped to fuel the rapid growth in deployment of intranets.

It didn't take computational chemistry software firms long to recognize the potential of web technology for computational chemistry. When the ACS exposition was held last August in Orlando, software companies had begun to implement strategies for using web technology to deliver computational chemistry tools to users. MSI, with its WebLab concept, and Tripos, St. Louis, with its Discovery.Net approach, are just two examples of the web-enabling activity now percolating through software development.

Bioinformatics a focus

As strong an influence on software development as web technology has been, that technology is still only an additional, albeit rapidly growing, mechanism for delivery of applications software. And on the applications front, bioinformatics is the hot new area of development. Bioinformatics is beginning to assume more of a role in the drug discovery process - another of the engines powering recent computational chemistry and information management software development.

Bioinformatics tools, computational chemistry software were among new products featured at ACS exposition booths, such as those of (from left) Tripos, MDL Information Systems, and Silicon Graphics.

Certainly the use of computational chemistry has been growing in other areas - polymers, catalysts, and materials in general. But with drug discovery energized by the entirely new technology of combinatorial chemistry and related techniques, computational chemistry software developers have been caught up in a race to tame that adjunct to pharmaceutical development.

Combinatorial chemistry generates lead compounds that could exhibit biological activity against a particular target. Since hundreds of thousands, even millions, of compounds can be involved, combinatorial chemistry is characterized by the vast amounts of data involved and the information management challenges they pose.

Addressing the information management needs of this process has kept computational chemistry and chemical information management software firms busy for several years now. But whatever success combinatorial chemistry is enjoying as a way to generate lead compounds, it says little about the biological target.

For drug hunters, helping to identify targets is a function of bioinformatics. That branch of computer-based information management deals with genomic information, from gene discovery to DNA and protein sequence, structure, and function - that is, the target itself. With both public and private genomic sequencing initiatives building up around the world, there is now a flood of information pouring into databases dealing not only with the human genome but with other genomes such as mouse and yeast as well.

"When you look at bioinformatics," says MSI's Savage, "that's a very hot area right now. It's also a very immature area." There are a lot of things happening, he says, with something new seemingly popping up on the landscape almost weekly.

A number of things popped up at the San Francisco exposition. Among those companies introducing new products were MSI; Oxford Molecular Group, Campbell, Calif., the U.S. arm of the company headquartered in Oxford, England; and MDL Information Systems, San Leandro, Calif., which currently is being acquired by Elsevier Science Inc., the U.S. subsidiary of the British/Dutch publishing firm Reed Elsevier (C&EN, March 31, page 7). Computer manufacturer Silicon Graphics, Mountain View, Calif., had demonstrations showing results of bioinformatics projects making use of the company's MineSet data mining software, which it introduced to the chemical community in a more general way last August in Orlando.

Molecular Simulations and Oxford Molecular Group also exhibited in San Francisco.


The products involved represent a mix of approaches to bioinformatics. In-house development, joint projects, and acquisition of marketing rights are all involved.

MDL, for instance, looked outside and acquired the marketing rights to two new products. In November, it signed an exclusive marketing agreement with Molecular Informatics, Santa Fe, N.M., to market that firm's BioMerge System, a family of bioinformatics software and databases that brings together public, third-party, and proprietary genomic data in a single, fully functional relational database management system. Then, in December, it acquired exclusive marketing rights from the National Center for Supercomputing Applications (NCSA) at the University of Illinois in Chicago for that organization's Bioinformatics Workbench. Bioinformatics Workbench is an intranet-based package that provides access to data and computing tools central to genomic research.

BioMerge has a number of integrated components. The basic element is BioMerge Foundation, which can be used to edit, browse, add, and query data, including physical map data and a variety of sequence and annotation information, from the chromosome level to a single base pair.

The other components augment BioMerge Foundation. Genomes Today is a DNA sequence database that provides on-demand updates of new sequences from worldwide public databases. MDL senior analyst William F. Hackett explains that in this way, a company's private data can reside together with the public data behind a company's own software security barrier, or firewall.

Bioinformatics Workbench has been called "point-and-click biology" by its developer, NCSA senior research scientist and University of Illinois professor Shankar Subramaniam. It integrates genomic databases and sequence-analysis packages under one simple interface on a corporation's own, secure intranet. Databases and analysis tools remain on the intranet server, Hackett points out, where they can be monitored and updated.

Oxford Molecular introduced its new Omiga software for analysis of DNA and protein sequences. Oxford notes that Omiga is the first in a family of integrated, enterprise-wide software tools and is available for Windows 95 and NT systems.

The company explains that Omiga can be used to identify biological targets for potential new drugs at a genomic and at a molecular level. For example, researchers can use the software to identify potential genetic targets for drug interactions by comparing and contrasting gene sequences with known and unknown activity. Gene cloning experiments can be simulated on the computer to look for likely outcomes. And expression of proteins of interest may be increased through analysis of DNA and RNA sequences.

For its part, MSI is combining in-house and partnership efforts in developing its software approaches to dealing with bioinformatics. In San Francisco, it launched WebLab Gene Explorer, which uses WebLab's browser-based computational environment for analysis of DNA and protein sequences and structures. Gene Explorer, Savage explains, is based on a concept the company is trying to introduce that it refers to as three-dimensional bioinformatics.

Gene Explorer interfaces MSI's structure analysis technology with the sequence analysis software of Genetics Computer Group (GCG), Madison, Wis. GCG, a private company originally founded in the department of genetics at the University of Wisconsin, Madison, has a sequence-analysis product called the Wisconsin Package, which contains more than 100 interrelated software programs.

The first module of the Gene Explorer suite is 3D Structure Prediction, which as the name implies transforms protein sequences into 3-D structural models. It provides all the necessary computational capabilities for sequence similarity searching, multiple sequence alignment, homology modeling, and structural verification. Gene Explorer will eventually include modules that provide functionalities for analysis of DNA sequences for restriction enzyme sites, DNA and protein homolog searches, 3-D protein sequence structure prediction, and graphical display of the predicted implications of amino acid mutations.

MSI also has a joint development program under way with Incyte Pharmaceuticals, Palo Alto, Calif., a company that designs, develops, and markets genomic database products, software tools, and related services. The joint project is aimed at integrating Incyte's LifeSeq gene sequence and expression database with MSI's Gene Explorer. The database contains information for more than 1.3 million human gene fragments derived from both normal and diseased cells and tissues.

Databases like LifeSeq present an enormous data management challenge. And indeed Incyte has been working with Silicon Graphics on ways to apply that company's MineSet package to bioinformatics, Silicon Graphics chemistry market manager David M. Zirl notes. Incyte is using the data mining tools, he says, to ask such questions as: "What is the information they have in there? Are there ways to better understand that data? Are there better ways to present the data?"

Silicon Graphics' MineSet data mining tools visualize data for a full-sequence comparison between genomes of Mycoplasma genitalium and Haemophilus influenzae. The data, generated at the European Bioinformatics Institute, are brought to life by an interactive display, where heights and colors of the bars indicate the relative similarity of each segment of the first genome to each segment of the second genome.


MineSet was first introduced by Silicon Graphics just a year ago. It is an integrated suite of tools for data access, data transformation, data mining, and visual data mining. Among its initial users were brand managers, production managers, market development managers, and data analysts trying to gain new insight into consumer, demographic, and industry data.

A Silicon Graphics demonstration at the San Francisco exposition exemplified the ways MineSet is being applied to bioinformatics. The demonstration visualized the partial results of a full-sequence comparison between two genomes, those of Mycoplasma genitalium and Haemophilus influenzae. Data for the demonstration were generated by Tom Flores, an R&D coordinator at the European Bioinformatics Institute near Cambridge, England, using genetic sequences available from the Microbial Database of the Institute for Genomic Research, Rockville, Md.

The data generated - the statistical results of biological sequence comparisons - required about 10,000 FASTA searches between the two genomes. FASTA is a widely used program developed by University of Virginia, Charlottesville, biochemistry professor William R. Pearson for making comparisons between protein or DNA sequences.

For the demonstration, Silicon Graphics used the Map Visualizer tool from MineSet to identify and locate multiple regions of similarity between the two genomes. The map presents either of the two genomes in a circular shape divided into equal segments, each representing a 1,000-nucleotide sequence in the genome. A slider represents segments of the H. influenzae genome, and moving it across the circular genome generates bars. Heights and colors of the bars indicate the relative similarity of each segment of the first genome to each segment of the second genome, with higher bars corresponding to greater measures of similarity.

Although a good deal of attention has been focused on bioinformatics, other new software is meant to be applied at different points along the drug discovery spectrum. The San Francisco exposition offered a look at a varied assortment.

One of these, DIVA, from Oxford Molecular, made its commercial debut in San Francisco. DIVA (for Diverse Information Visualization & Analysis) is designed to provide the tools for rapidly and flexibly dealing with large sets of data such as those of, say, the 2,000 to 200,000 or more entries now common in pharmaceutical research. It can be used in such application areas as high-throughput screening, combinatorial chemistry, and medicinal chemistry.

The San Francisco exposition marked a complete launch of the product, an early version of which Oxford Molecular previewed in Orlando. DIVA's origin is perhaps unique, in that it grew from a collaboration between engineers from Oxford Molecular and scientists from pharmaceutical manufacturer Glaxo Wellcome, working together at Glaxo Wellcome's research facility in Research Triangle Park, N.C. The two companies had entered into an agreement, now up for renewal, to codevelop a fully integrated and expandable software system aimed at expediting Glaxo Wellcome's drug discovery process and to transfer the developed software to Oxford Molecular's commercial product line.

DIVA is designed to be extremely flexible. It can import data from local or remote sources. It can retrieve structures and perform substructure searches. It can generate different views of the data, such as a structure-oriented view or a data-oriented table view, and can minimize a view to examine large quantities of data. It can manipulate the view in many different ways, such as color coding of table cells to observe large-scale trends. It can perform detailed analyses, such as histograms of continuous or noncontinuous data, as well as FIRM (Formal Inference-based Recursive Modeling) analyses, a form of decision-tree analysis where a large population is split into subgroups based on important predictor variables. And it can plot 2-D and 3-D graphs.

Tripos, a computational chemistry software firm that focuses essentially on pharmaceuticals and biotechnology, has a constellation of new initiatives under way for application to drug discovery research. The large data sets stemming from high-throughput screening and combinatorial chemical synthesis are the focus of one of these initiatives, ChemEnlighten. ChemEnlighten is a new web-based software package Tripos introduced in San Francisco for use in performing statistical studies of chemical-structure data sets of virtually any size.

A broad-based application tool for data filtering, ChemEnlighten is designed to manage these large sets of chemical structures and any number of related property calculations. It includes metrics engines that are used to calculate properties on the data sets. These include molecular weight, molecular volume, log p (octanol partition coefficient), Tripos' Unity 2-D fingerprint, and more. Unity is a query and analysis system for searching multiple, distributed databases of compounds.

A significant feature of ChemEnlighten is that it includes an optional interface to Molconn-Z. Molconn-Z is a standard program developed by Lowell H. Hall, a chemistry professor at Eastern Nazarene College and head of Hall Associates Consulting, both of Quincy, Mass. It generates indexes for molecular connectivity, shape, and electrotopological state (E-state).

Tripos sees Molconn indexes as an important addition for ChemEnlighten customers wishing to perform molecular diversity analyses on hundreds of thousands of compounds, according to Trevor W. Heritage, director of product marketing at Tripos. The indexes, he says, have been shown to be important as a basis for molecular similarity comparisons and for chemical database characterization, including similarity searching and quantitative structure-activity relationship (QSAR) analyses.

QSAR is also the target for another of Tripos' initiatives, Hologram QSAR (HQSAR), a new software tool due out this summer. HQSAR is a structure-activity prediction tool designed to predict biological activity and other properties from a series of compounds with varying activity. It uses a hologram - a special electronic fingerprint based on molecular fragments - to guide the prediction. HQSAR doesn't replace conventional 3-D QSAR techniques but is complementary to them.

"It's very, very, very fast," says Heritage. A QSAR study of a set of 40 to 50 molecules might take 10 minutes to perform, he explains, whereas a typical study using current techniques may take two or three weeks to get the alignments right. "So it's a very fast technique for having a quick look at your data," he adds.

Yet another of Tripos' initiatives is a new collaboration it has begun with chemistry professor Norman L. Allinger, director of the Computational Center for Molecular Structure & Design at the University of Georgia, Athens. Tripos has been the exclusive distributor for MM3, a leading molecular mechanics program for small molecules designed and developed by Allinger.

The collaboration is aimed at extending the MM3 program so that it can handle large molecules and the interactions between large and small molecules. Tripos has found that many of its customers want to perform docking studies and studies of small molecules in the context of the biological receptor, Heritage explains. But, he says, there is no force field - the underlying information and equations characterizing a particular type of molecule - that handles both small and large molecules simultaneously. "It makes their studies very difficult, and they end up making approximations that they're not happy with," Heritage says.

Chemical Design Ltd., a U.K. computational chemistry software firm with U.S. offices in Mahwah, N.J., has launched a new module for its Chem-X computational chemistry software suite. Called ChemIdea, it is directed toward lead compound optimization.

ChemIdea is designed to suggest appropriate substituents to include in new molecules by exploiting existing data from biological activity testing. It uses test results from both active and inactive molecules to produce clear volume/ hydrophobicity plots and critical molecular properties for drug design.

These plots enable chemists to identify areas of unexplored property space and explore them in terms of new substituents that possess the desired properties. Floating the cursor across the plot on the computer screen causes possible new molecular fragments to display on the fly. The sources can be commercial catalogs, in-house reagent databases, or a combination of both.


SPOTLIGHT SWINGS TO EDUCATION


Exploiting web technology

MSI's Genome Explorer and Tripos' ChemEnlighten packages are just two examples of the software being developed that exploit web browser technology. Other new web-minded programs surfaced at the San Francisco exposition as well.

MSI, for example, introduced the first module of another component in its WebLab strategy, WebLab Polymer Explorer. The module is called Property Predictor, and it draws on statistical correlation methods implemented in the company's core computational chemistry program, Cerius2. Quantitative structure-property relationships (QSPR) are used to calculate mechanical, thermophysical, and transport properties of amorphous linear polymers. The module guides users step-by-step through the QSPR calculation. For polymer chemists, it provides a way to screen candidate polymers for particular applications before synthesizing them and to determine the best copolymer ratios. Formulation chemists can optimize the properties of blends and mixtures.

MSI also announced the release of version 2.0 of its WebLab Viewer, the basic molecular visualization package for such WebLab applications. Among other enhancements, the new version has improved visualization for molecular surfaces, proteins, and crystalline solids; improved annotation and labeling capabilities; and the ability to display hydrogen bonds and hydrophobicity or charge.

Since early this year, Advanced Chemistry Development (ACD), Toronto, has been developing a unique activity on the World Wide Web called ACD/ILab (Interactive lab). At the company's web site (http://www.acdlabs.com), Java-based applications allow researchers with Netscape 3.0 to search databases and predict chemical properties. Such Java applets - small programs provided by a server that download within a web page for short-lived use - are a way of augmenting web technology. Java is a computer language developed by Sun Microsystems for writing programs to run on the web within a browser.

Advanced Chemistry Development uses Java applets strategy.


ACD has a growing family of products for structure drawing and property prediction. ACD/ILab is based on many of these products. Through the Java applets, it enables users to draw structures and, by clicking, calculate different properties for the structures. For example, it provides the IUPAC (International Union of Pure & Applied Chemistry) name, boiling point, vapor pressure, heat of vaporization, log p, and pKa. It also calculates and displays accurate carbon and hydrogen nuclear magnetic resonance spectra and enables users to manipulate them. The interactive lab has been a free service during development, but the company plans to begin assessing a per-use charge soon.

All of this represents a new direction of chemical information management for the company. According to ACD sales representative Paulius Jurgutis, the company's focus is to lead Internet development of chemical information through Java applets. Indeed, it is now developing a web-based laboratory information management system (LIMS).

Java applets are also the basis for a new product from Cherwell Scientific Publishing, Oxford, England. The product, ChemSymphony, is a collection of some 28 Java applets that allow interactive 3-D molecular structures to be easily incorporated into documents encoded in hypertext markup language (HTML) for web publishing. The user can select the features needed and the level of interactivity desired for a viewer. The applets then make it possible for a viewer to rotate, translate, and interact with the molecular structures.

A related development involves MDL Information Systems' Chemscape Chime, a Netscape plug-in that enables users to access manipulable chemical structures through web browsers. Plug-ins extend the basic capabilities of browsers - for example, enabling a user to manipulate graphics - within the browser window itself. Interactive Simulations, San Diego, has a product called Sculpt, a molecular modeling program that lets a user interactively twist, tug, and tether molecules directly, while atoms move to accommodate the changes. In a collaborative effort between the companies, the sculpting ability of Sculpt has been added to the Chime plug-in.

Although web technology along with drug discovery as focal points of development have grabbed the headlines over the past couple of years, development activity on core computational chemistry packages has continued. The San Francisco exposition saw a wave of new offerings in this area, with the latest versions of established flagship programs making an appearance. But visitors to the exposition also found something conceptually new.

That something is Molecular Operating Environment (MOE), developed by Chemical Computing Group, Montreal. The company is new, and the product has been on the market only since early this year.

MOE is what Chemical Computing Group marketing vice president Bill Hayden calls "computational chemistry with a difference." It is not a software package in the usual sense. Rather, it was conceived as both a general-purpose chemical computing environment to facilitate all forms of computational chemistry and as a methodology development platform, such that chemists can build and assemble their own tools quickly with the built-in high-performance programming language.

Paul Labute, Chemical Computing Group's director of R&D, believes that for industrial research companies, in-house methodology development capabilities will be of fundamental importance to their long-term competitiveness. But that presents a problem. If a researcher publishes an interesting and possibly useful new methodology, Labute reasons, it probably won't be quickly incorporated into existing commercial computational chemistry software.

"We wanted to let people spot something they like, encode it very quickly, test it out, perhaps expand upon it, add their own ideas, and play with the algorithms that come out all the time," Labute says. Indeed, a survey by Chemical Computing Group of several hundred senior computational chemists found strong support for the concept, he adds.

But for this use, Labute says, traditional general-purpose programming languages such as C and Fortran come up short. So Chemical Computing Group invented a new language, which it calls Scientific Vector Language. SVL is a high-performance, data-parallel language embedded in MOE. That is, its compiler and run-time environment are an integral part of MOE, and SVL serves as the command, macro, scripting, and high-performance computing language of MOE.

For conformational space exploration, the basic MOE system provides a collection of SVL programs for use as is or as a basis for further methodology development. Groups of application programs are also available in the areas of molecular similarity, biopolymer applications, and electrostatics and miscellaneous applications. Molecular Computing Group doesn't plan to develop its own chemical information systems but has short-term R&D plans for including SVL services to connect to servers from both MDL and Daylight Chemical Information Systems, Mission Viejo, Calif.

MOE currently runs on Silicon Graphics, IBM, and Sun Microsystems workstations, and is in the beta test stage on Intel Pentium processor PCs running Windows NT. MOE will also be ported to Silicon Graphics shared-memory multiprocessors, Cray computers, IBM SP2 parallel computers, and Hewlett-Packard UNIX workstations. Molecular Computing Group's goal, Labute says, is to have every SVL program automatically "parallelized," which would mean that a single SVL source program would be used for both sequential and parallel computers.

Another novelty that visitors to the San Francisco exposition saw is a new name for the PS-GVB software package developed by Portland-based Schrödinger Inc.: Jaguar. The name change represents the culmination of six years of development that have brought the product to a point of completion. The new name, evoking Austrian physicist Erwin Schrödinger's cat, was chosen, according to the company's president and chief executive officer, Murco Ringnalda, to reflect the speed and power of the software package.

Goddard points to Schrödinger's display sporting a new name for its software.


Schrödinger the company was founded in 1990 by chemistry professors William A. Goddard III of California Institute of Technology and Richard A. Friesner of Columbia University. Goddard recalls that they outlined a plan for what needed to be in a software package for quantum mechanics to be fast enough to be a useful tool to solve real problems rather than just providing increased understanding. With the release of version 3.0 of Jaguar, that plan has now been completed, he says.

"It turned out that the problem was that the accuracy to give an answer where you didn't need to do an experiment wasn't there," Goddard says. "It wasn't in the force fields. That is, the springs that were used to represent the molecules didn't have enough information. It's really the quantum mechanics that Schrödinger's been developing that now finally has the accuracy that you can use it to replace an experiment."

For moderate-size systems, Goddard claims, the new methods are about 15 times faster than traditional methods and they scale better. As the next generation of systems comes along, ones that are, say, 10 times larger, they will probably be around 150 times faster, he says, adding: "That's definitely a quantum leap forward."

Jaguar is an ab initio electronic structure software package that includes standard methods such as Hartree-Fock, density-functional theory (DFT), and second-order Møller-Plesset perturbation theory (MP2), as well as novel approaches such as generalized valence bond (GVB), localized MP2 (LMP2), and GVB-LMP2. To solve the equations of electronic structure theory, Jaguar employs a unique algorithm that is a synthesis of conventional quantum chemical techniques with a modified version of the pseudospectral (PS) method, a numerical approach widely used in hydrodynamic simulations. Jaguar's former name, PS-GVB, represented the original methodology.

To push the next step in development, Schrödinger is forming a partnership that includes industrial partners. Novartis, the Swiss pharmaceutical manufacturer resulting from the merger of Ciba-Geigy and Sandoz, is already on board, and another pharmaceutical manufacturer is all but signed up. A number of academic partners in addition to Goddard and Friesner complete the partnership. The idea, Goddard explains, is that the academic partners will provide a basic research component, Schrödinger will provide the software development component, and the industrial partners will provide the drive toward real problems.

Visitors to the ACS exposition in Orlando last August had a chance to preview another then-new quantum chemistry program - one from what was then also a new company, Q-Chem Inc., Pittsburgh. Version 1.0 of that program, Q-Chem, had its commercial introduction at last month's San Francisco exposition.

Q-Chem is an integrated ab initio software package incorporating novel quantum mechanics technology. A main inducement for the development of Q-Chem was to implement the continuous fast multipole method for calculating Coulomb energy, in turn providing Q-Chem with linear scaling for the DFT method. The Q-Chem package now also incorporates new methods that provide it with linear scaling for the Hartree-Fock method. So it's possible to operate with linear scaling for the hybrid combination of the two - something no other program can do, according to Q-Chem President Benny G. Johnson.

Its capabilities enable Q-Chem to address molecular structures larger than those that can normally be handled in a feasible amount of time. An example is the calculation of a 352-atom conotoxin peptide on a 32-processor Cray T3E supercomputer that took 30 minutes. Conventional technology, Johnson says, would require an hour to calculate a molecule of about 60 atoms.

The 1.0 release of Q-Chem is available united with the graphical user interface of HyperChem. HyperChem is the desktop molecular modeling package of Hypercube, Waterloo, Ontario. Hypercube announced at the San Francisco exposition that it had opened a new office in Gainesville, Fla., which it expects will eventually become the principal business office for the company, although the headquarters will remain in Waterloo. The new Chemist's Development Kit of Release 5 of HyperChem, launched last August, was used to define a graphical user interface for Q-Chem.

Q-Chem Inc., is also working with Oxford Molecular to combine that company's UniChem molecular modeling package as a graphical user interface with Q-Chem. The companies expect the product to be ready for release in the second quarter of this year.

The new editions of core computational chemistry packages further extend the capabilities of those systems. The following are among the ones introduced in San Francisco:

Chemical drawing programs have not remained static either. CambridgeSoft, Cambridge, Mass., introduced in San Francisco a version 4.0 of CS ChemDraw, a widely used structure drawing program. At the same time, the company has added an Ultra product category to the existing STD and PRO categories for its CS ChemOffice suite, which provides an integrated desktop environment for structure drawing, molecular modeling, and chemical information management.

Michael J. McManus, CambridgeSoft director of marketing, notes that chemical intelligence is one of the big additions to ChemDraw. For example, it automatically cleans up structures, converts atom labels from text to actual structures, checks for chemical syntax errors, adds bonds to any text character in an atom label, draws organometallics, and draws mixtures of isomers with a single structure. It also eliminates the need for hand calculations by computing molecular formula, molecular weight, exact mass, and elemental composition.

In CambridgeSoft's new product groupings, ChemOffice STD pairs ChemDraw Pro and Chem3D Pro for structure drawing and 3-D visualization and computation. ChemOffice Pro adds Mopac Pro, which integrates the semiempirical Mopac 93 computational chemistry program of Japan's Fujitsu Ltd. with Chem3D's graphical interface. The Pro package also includes ChemFinder Pro, a chemical database application. The new ChemOffice Ultra 4.0 expands the Pro package further. It includes ChemInfoPro, a collection of databases for ChemFinder and ClipArt Pro, for adding drawings of equipment and apparatus to presentations or publications.

Meanwhile, a variety of other new and anticipated software offerings are and will be adding to the chemist's computer-based armamentarium.

One comes from Synopsys Scientific Systems, Leeds, England, a provider of computerized scientific information products and services for pharmaceuticals, fine chemicals, and biotechnology. Its Accord for Microsoft's Excel spreadsheet software, for example, turns Excel into a chemical spreadsheet. The company has now released version 2.5 of the Accord Software Development Kit, a 32-bit edition, making it easier to develop chemically aware applications for Macintosh and Windows 95 and NT. The new version provides support for storage, searching, and display of structures and reactions, as well as other chemical representation enhancements.

Microsoft Excel also figures in the plans of Beilstein Information Systems, a Frankfurt-based firm with U.S. offices in Englewood, Colo. In excerpting the chemical literature from 1779 until now, Beilstein has amassed a database in organic chemistry containing chemical structures and their associated chemical properties, physical properties, preparative methods, chemical behavior, and literature references for more than 6 million compounds. Beilstein's CrossFire client/server system enables users to search the database files from PCs.

Beilstein is now working on an adaptation of Excel as a front end for searches by CrossFire, according to Jorge Manrique, director of marketing for Beilstein in North and South America. Retrieving Beilstein data into an Excel spreadsheet that is set up the way a user wants will effectively provide the user with a miniature chemical database, Manrique adds. He expects a formal product release soon.

SciVision, Lexington, Mass., is a developer of third-party applications for use on PCs. It has a SciLogP product, for example, for use in drug design, and a SciPolymer program for use in de novo design of new polymers.

Joseph R. Votano, SciVision president, says the company is now working on a program that takes a neural network approach to calculating log p. Linear regression has been used, but the process of calculating log p is really nonlinear, Votano points out. The neural network approach takes that nonlinearity into account. Validation testing has been done on the program, Votano says, and it may be out by this summer. Also on tap for late summer release is a new topological program for use in drug research. To be called SciMolconn-X, it will contain more than 250 topological descriptors for doing QSAR analysis.

Late summer will also probably be the release time for a new edition of SciVision's SciGlass program, an integrated system for glass information and property calculation. The new version will contain information on 80,000 glasses and 700,000 properties. It will also have several new property prediction algorithms beyond the 20 in the current version.

Altogether, the computational chemistry software developments on view at the San Francisco exposition underscore the breadth and depth of a technology now moving sure-footedly into the mainstream.




ChemCenter Pubs Div. Home Page
Copyright © 1997 by the American Chemical Society.