Laboratory Technology Trends: Information Technology in the Life Sciences I.T.'s INTENSIFYING INFLUENCE Current research in life science would be almost impossible without the tools of information technology. New hardware and software simplify the tasks of collecting, archiving, and analyzing the complex data that experiments generate and permit investigators to make reliable comparisons among information stored in widely different formats.

BY PETER GWYNNE AND GARY HEEBNER

ADVERTISERS InforMax, Inc. [USA] Bioinformatics software solutions for the analysis and interpretation of genomic, proteomic and other biomolecular data 240-747-4000 www.informaxinc.com InforMax, Inc. [UK] +44 (0) 1865 784 580 ISI Thomson Scientific [USA] Web-based and printed information providing coverage of the scientific publications and other relevant research. 215-386-0100 www.isinet.com ISI Thomson Scientific [UK] +44 (0) 1895 270 016 Messe München GmbH [Germany] Services for exhibitors and organizes trade show and scientific conferences +49 89 9 492 07 20 www.messe-muenchen.de

CONTENTS • Enter Information Technology • Two Roles for IT • Compatibility and Standards • The Arrival of Supercomputers • Beyond the Hardware • Simple Statistical Analysis • Software for Sequencing... • ...and for Proteomics • The Chemists' Perspective • A New Paradigm • Identifying Drug Targets • The European Example

The companies in this article were selected at random. Their inclusion in this article does not indicate endorsement by either AAAS or Science, nor is it meant to imply that their products or services are superior to those of other companies.

» You arrive at the laboratory in the early morning expecting to find an important e-mail message from a colleague. But when you open it, the message contains only a senseless garble in which words run together without punctuation. Rather than trying to decipher the contents, you decide that it's easier to call your colleague and ask her to fax the message.

» What caused the problem? A mismatch between two word processing programs might have been responsible. So might the failure of the sending and receiving e-mail servers to communicate properly. What is clear is that your e-mail system has welcomed you to the world of electronic incompatibility.

» Scrambled e-mail messages represent only a minor hindrance to the work of the modern life science laboratory. Far more threatening is the issue of incompatible information. Experimental data today can exist in many varied formats and reside on several different platforms. "No one database or system can pull everything together," explains Mike Colvin, computational biology group leader in the Lawrence Livermore National Laboratory's biology and biological research program. The inability to convert information from one database into the format of another can make it impossible to compare the results of similar experiments in different laboratories that use different database technology. Since a major challenge for present-day life science researchers is to determine relationships between data in different formats and platforms, that inability looms as a significant roadblock to progress.

Exacerbating the problem is the sheer volume of information generated by academic and industrial biology labs. Genome sequencing and postgenomic efforts to delve into proteins and protein pathways have made life science an extraordinarily data-intensive pursuit. "only high energy physics and perhaps astronomy generate as much data," says Burton Smith, chief scientist of Cray, Inc. Few experiments yield yes/no answers. Instead, researchers must mine deep into the information produced by individual experiments and series of runs to understand the implications of their work and to compare their results with those from other experiments in their own and other laboratories. "You can't do anything in the life sciences without generating a lot of information," says Robert Kehrer, partnership manager, worldwide developments for Apple.

ENTER INFORMATION TECHNOLOGY

This is the point at which life science encounters information technology (IT). "IT and computing will be essential to life science," says Colvin. "The life sciences are information science now." Jim Lindelien, CEO of TimeLogic Corporation, agrees. "IT and computer technologies are critical to the next advances in life science research," he says. "There seems to be a never-ending stream of demands on computing resources — through all genome applications and beyond that, into proteomics, and on up to the drug development part of the research pipeline."

Eli Mintz, president and cofounder of Compugen, expands on that thought. "Just as you can't do modern physics research without IT, you can't do life science research without it anymore. It would take humans forever to do the necessary calculations. New ways to analyze the data are becoming very basic to genomics and proteomics."

Computers have arrived in the life science lab with remarkable speed. "When I was a student in the mid 1980s," recalls Jeff Augen, director of strategy for IBM Life Sciences, "it was possible to complete an entire thesis project without ever using a computer. Today, if the computers are broken the research stops." In fact, he continues, "IT is the core value in the life sciences today. Without IT, nothing happens."

To make things happen, vendors now market several varieties of hardware, software, database technologies, and other computing accoutrements aimed specifically at the biology research market. The vendors aim to make computing accessible to life scientists who, while they may have more experience with IT than their predecessors in the lab, still lack the deep and detailed skills of computer scientists. "User friendliness is one aspect of what we do as tool developers and manufacturers," explains Mintz. "We expect that biologists will eventually understand what computer sciences can contribute to life science. But at present I'm not sure that their curriculum has caught up with the changes that IT has brought to biology."

Some software programs for data handling in the life science laboratory are very general in nature, used for routine spreadsheet and statistical calculations. Other forms of software and hardware are designed specifically for application to genomics, proteomics, and drug discovery. Several companies are also working on tools to improve the connectivity of data generated in different laboratories and with different software programs.

TWO ROLES FOR IT

According to Colvin of Lawrence Livermore, information technology has two main roles in life science. First comes the combination of data collection and archiving. That segment, he says, "is big and getting bigger." The numbers tell the tale. "If you store all the data coming off sequencing machines you have 300 terabytes," IBM's Augen points out. "Storing all the trace files for the human genome project would require 300 terabytes — an enormous expense. Some day we will sequence the genome of each individual. The costs associated with storing this data would be astronomical. Today, for financial reasons, Celera keeps approximately three terabytes of the original data." Postgenomic work promises to create even more information. "People doing mass spectrometry for protein sequencing are generating 250 gigabytes per day per machine," Augen continues. "Some companies are generating 20 terabytes per day."

Also growing in importance is the use of IT for quantitative simulations of biological processes, which range from modeling of chemical processes and of proteins all the way up to predictive modeling of cellular pathways and the effects of drugs on tissues. "Simulation exists primarily in research labs in academe and the national laboratories," says Colvin. "But a rapidly growing fraction of papers refers to a model. Life science is essentially following a curve taken by chemistry a decade ago. The challenge lies in the fact that the processes are much more complex."

Life science must also follow a curve taken by the computer industry. Two decades ago, the business relied largely on proprietary technology. Computers, operating systems, and databases from different vendors could not communicate with each other. But once industry leaders realized that this system limited customers' ability to use computers, and hence their demand for the industry's products, they started a movement toward openness. That goal remains elusive, as the example of incompatible e-mail illustrates. But the advent of operating systems such as Unix and Linux has permitted computers from different vendors to communicate understandably with each other.

COMPATIBILITY AND STANDARDS

In life science, the issue of compatibility focuses mainly on databases. Not only do different laboratories store their burgeoning biological information in different formats. The nature of the data also varies significantly from one database to the next. So does the reliability of stored information. Thus the IT industry is seeking appropriate responses. "There will be a place for communication tools that are interfaces to different types of data — a kind of toolkit for interfacing that will make the system transparent to individual life scientists," says Colvin. "There's also a need to ensure that the databases don't get polluted by less solid data. A lot of this, I think, will depend on an open-source, Linux-type approach."

Efforts have already started to develop standards for biological databases. The Interoperable Informatics Infrastructure Consortium (I3C), an organization headed by the Biotechnology Industry Organization (BIO), plans to create a standardized global language for the biotechnology and pharmaceutical industries. The consortium intends to help establish the types of biological information that databases can contain and to recommend a set of rules for searching, manipulating, and linking the data.

Another nonprofit organization, called blueprint WORLDWIDE INC. and founded by IBM and MDS Proteomics, has taken its own step to creating a standard. Last May it unveiled the world's first public Biomolecular Interaction Network Database (BIND) as a comprehensive source of protein interactions that trigger chemical reactions responsible for diseased or healthy cells in the body. "We're now involved in gathering funds to hire people to get data into BIND," says Francis Ouelette, director of the Centre for Molecular Medicine and Therapeutics Bioinformatics Core Facility in Vancouver, Canada, and blueprint WORLDWIDE's managing director designate.

THE ARRIVAL OF SUPERCOMPUTERS

The application of IT to life science starts with hardware. Several companies are now developing the hardware and operating systems required for storing and manipulating the vast amounts of data generated by researchers in the life sciences. They are also working to translate users' requirements into IT solutions in the laboratory.

As life scientists have collected more data of greater complexity, they have started to apply physical scientists' hardware of choice — supercomputers — to their analyses. These formidable tools, loosely defined as the most powerful class of computers available at any point in time on the basis of the number of floating point operations (FLOPs) they can perform each second, can solve large, complex problems that are too challenging for less powerful computers. Since the first Cray machine arrived in the mid 1970s, supercomputers have contributed enormously to the advance of knowledge and the quality of human life. They have created new materials. They have powered advances in electronics and visualization. And they have begun to unravel such mysteries as the shape of the universe.

Now, supercomputers face a fresh challenge. "So far they have not played a really big role in the life sciences," says Cray's Smith. "Most of the performance needs have been met by computer farms that can distribute the work to smaller machines. But that is changing. People are now interested in using supercomputers for sequence matching and searching problems. Various kinds of special purpose hardware have been built to pursue some of those tasks."

Cray recently provided an SV1 supercomputer to the National Cancer Institute for a genomic sequencing project. The machine, a parallel computer with a theoretical processing speed of 115 gigaFLOPS, has potential applications in protein folding, one of the most data-intensive problems in life science, and more. "It might be used to predict protein structures by going beyond protein folding to getting the detailed structure of the docking site for a drug and the grosser structure of the protein," Smith explains.

IBM, meanwhile, is developing its own supercomputer for life science. Called Blue Gene, it is 500 times more powerful than the world's fastest machines and capable of more than one quadrillion floating point operations per second (one petaFLOP). "The Blue Gene project helps us learn about advanced computer architectures with self-healing characteristics — an absolute requirement for systems in the petaFLOP range," says Augen. "Protein folding is a perfect application for such a system. Our computational biology research center has around 80 to 90 scientists with backgrounds in nuclear magnetic resonance, X-ray crystallography, and the general application of computer science to the life sciences."

BEYOND THE HARDWARE

Other prominent computer manufacturers, including Apple, Compaq, and Sun Microsystems, have started to develop instrumentation that will help the life science community to manage its information overload. Like IBM, Apple has formed a separate division to focus on developing tools and technologies that will improve life science researchers' productivity and data analysis.

The efforts go beyond new hardware. Apple, for example, has developed the Mac OS X operating system for life science researchers. This industrial strength system provides the multitasking, memory-protected environment needed for scientific exploration while retaining the ease of use that scientists require to be productive. Mac OS X delivers an intuitive and visually stunning interface. It also allows users to access Unix commands through a terminal window. "Life scientists have typically had two machines on their desks — a powerful Unix machine and a PC," says Richard Kerris, Apple's director of development technologies. "Now for the first time they have been able to combine those needs in one machine. The life science community is very Mac favorable because a lot of the familiar tools such as scripting languages from Unix are starting to pop up on OS X. And our Macs can talk to supercomputers as effectively as any other Unix box."

Like other vendors, Apple also works on the interface between hardware and software. For example, it has enabled TurboBLAST from TurboGenomics. This software is a cross-platform, accelerated, parallel implementation of NCBI's BLAST application. TurboBLAST allows scientists to harness the idle cycles of their desktop machines or to use dedicated clusters to create an affordable in-house BLAST solution with the power to handle even the most complex tasks. By leveraging the power of Motorola's G3 and G4 processors and Apple's Mac OS X, TurboBLAST delivers supercomputer performance at a fraction of the cost.

TimeLogic, meanwhile, aims to help life scientists make the most of their IT resources through a process called reconfigurable computing. "We're in the business of producing very high-performance hardware to leverage the value of critical bioinformatics software," explains Lindelien. "We introduce a hardware accelerator into conventional servers and choose to accelerate those algorithms that create bottlenecks or are too costly for the organization to solve in a reasonable amount of time. We get more bioinformatics analysis capacity at one-tenth to one-hundredth the cost of adding additional central processor units to a server farm."

SIMPLE STATISTICAL ANALYSIS

While hardware provides the computing power, software gives life scientists the opportunity to tailor information technology to their needs. Those needs range from the general — word processing, spreadsheets, and statistical analysis, for example — to the highly specific, among them interpreting DNA sequencing in genomic research and carrying out molecular modeling in drug discovery.

The general programs exemplify the increase in sophistication of life science experiments. "The old saying — if you need to analyze your results you've done the wrong experiments — still holds in some situations," says Harvey Motulsky, president of GraphPad Software. "But with many experimental systems statistical analysis is essential."

Even routine statistical analysis can take several hours to perform in the absence of a computer. The everyday calculations are fairly tedious and mistakes can be difficult to find and correct. Several companies have developed user-friendly programs that simplify this work. SAS Institute, among other firms, has programs for basic calculations as well as scientific graphing and curve-fitting. The software ranges from basic packages to complete families of products.

To make statistical calculations accessible to scientists who don't need a heavy duty statistical analysis product, GraphPad has created InStat, designed for any scientist to master in a few minutes. The company has also created a larger program, Prism, that combines making scientific graphs, doing basic laboratory statistics, and fitting curves. Unlike other data analysis programs, says Motulsky. "Graph doesn't assume you already know a lot about statistics. Not only are the programs easy to learn and use, but they also guide you to pick an appropriate test and help you interpret the results."

SOFTWARE FOR SEQUENCING...

The software demands of simple statistics pale against those of major projects such as sequencing the genomes of humans and other organisms. The generation of vast amounts of scientific data in these efforts has driven the formation and fueled the growth of bioinformatics, the field that focuses on drawing rational conclusions from information too voluminous and too scattered to be analyzed by a single scientist or research team.

A growing number of companies offers software for analyzing DNA sequences and protein structures. These products and services often include access to proprietary databases with large volumes of sequence data. Some systems can be accessed through the Internet, a route that allows researchers to manipulate large data sets without having to make extensive investments in PC hardware. Others reside in the user's facility, offering greater security. Companies that offer suites of bioinformatics programs include Biomax Informatics AG, Entigen, and InforMax.

Europe has a strong presence in this market. One of the newest organizations to emerge in bioinformatics is German firm Axaron Bioscience. Founded last November in a joint venture between BASF and LYNX Bioscience AG, the company specializes in functional genomics. It integrates scientific expertise and technological know-how in functional genomics to provide solutions that increase the quality of life. Other European firms that have become major providers of software and services in the bioinformatics area include the European Bioinformatics Institute, GeneBio, Genomatix Software, and LION Bioscience.

In the United States, Compugen has pioneered computational applications in genomics and proteomics. "Our expertise is in data mining," says Mintz. "We have developed the tools to analyze data currently available — mostly sequence data." Compugen's approach combines mathematics and computer science with molecular biology to improve the understanding of genomics and proteomics. "We have designed a set of algorithms, the leads platform, that can handle data and databases of all the genes," Mintz explains. "We analyze them using a complex modeling phenomenon that models, say, alternative splicing, contaminations, and other artifacts."

Mintz emphasizes the dangers of oversimplification in analyzing the results of genomics experiments. "If you oversimplify you don't get good results," he explains. "We have made sure that we don't oversimplify."

...AND FOR PROTEOMICS

Proteomics presents problems of analysis even greater than those encountered in genomics. In response, new software has emerged for analyzing peptide sequences and two-dimensional and three-dimensional protein structures. Protein identification and characterization typically involve the use of two-dimensional gel electrophoresis and mass spectrometry.

"Our V3 is a breakthrough technology that can compare two-dimensional gels and find the spots and the differences between different gels," says Mintz. "The advance involved the exact alignment of two gels. It will give 2-D gels a new push. You would invest in it if you have a lot of 2-D gel work."

Several other companies, among them Amersham Biosciences, Applied Biosystems, Bio-Rad Laboratories, and Sigma-Aldrich, have developed families of instruments, reagents, and software for proteomics research. Some vendors offer software for protein analysis along with their mass spectrometers. "Without networked computing technology and sophisticated informatics software, most high throughput techniques that involve mass spectrometirc detection would be unworkable or commercially unviable," says Mark McDowell, marketing director for Micromass UK. "The number of spectra of merit recorded in a typical proteomic analysis is so large that it becomes prohibitively expensive to interpret them manually."

Micromass, which develops mass spectrometers (MS) and MS software, has partnered with other firms to provide integrated solutions for proteomics research. It has, for instance, worked with Bio-Rad to produce the Bio-Rad ProteomeWorks System. "We realized about 10 years ago, when many MS platforms were running on relatively expensive minicomputer platforms, that the way to go was knowledge generation with PCs," McDowell recalls. "Our novel Mass-Informatics strategy was to change the concept of computing in mass spectrometry. Traditional data systems simply recorded, stored, and displayed limited volumes of MS data. Our approach is founded on the automated interpretation of high resolution MS data — turning samples into knowledge and not just spectra. We asked protein scientists to describe their challenges and priorities so we could translate them into 'high level MS language.' As a result we have developed an accessible interface to the power of mass spectrometry, enabling life scientists to rapidly map the protein landscape."

Other companies are developing proteomics software programs for their own in-house use. Several will partner with other organizations or will license their software to others. Such companies as GeneProt, MDS Proteomics, and Oxford GlycoSciences are taking this approach.

THE CHEMISTS' PERSPECTIVE

The biological form of informatics isn't the only one to rely intensely on IT. Chemists working in high throughput screening and similar applications would struggle to categorize, archive, track, and locate the large numbers of compounds they need for their work in the absence of software products from companies such as MDL Information Systems and Tripos that specialize in IT for cheminformatics. These programs feature such capabilities as chemical structure searching and information integration. They permit a scientist to work with many attributes of chemical compounds even if they exist in different formats.

CambridgeSoft Corporation has a strong position in this area as a supplier of software, solutions, and services based on Internet browsers and web servers for chemistry. "We have a very broad array of tools from desktop to enterprise for the work processes of our customers," says vice president of research and development Michael Rubenstein. "The workflow of organizations has evolved in such a way that they all need to share data and communicate. In addition, the work flows across scientific disciplines. Our role is to facilitate the information sharing."

The need for sharing is particularly evident in large organizations. "They have difficulties with deployment of software that limit their ability to incorporate new technologies," says Rubenstein. "That's particularly true when products have to be installed across all the desktops in the organization. Those problems are getting bigger as corporations sign up for research collaborations."

CambridgeSoft's ChemOffice WebServer is an enterprise implementation of ChemOffice that facilitates applications based on browsers and web servers. "In terms of internal structure our business has two significant foci: desktop tools and enterprise solutions," says Rubenstein. "Today we see a much greater emphasis on enterprise solutions. Even academic laboratories have them." The ChemOffice suite includes the ChemDraw program used by many chemists.

A NEW PARADIGM

IT's increasing influence on life science promises most benefit to drug discovery. Early efforts in this field involved the slow, labor intensive process of screening natural products derived from plants and microorganisms and testing them for activity in animal models. The past decade's advances in molecular biology, genomics, automation, detection, and informatics have created a new paradigm. The fresh approach to drug discovery and drug design relies heavily on computational power, and shifts the scientist's efforts from basic laboratory research to virtual research in silico. In particular, the approach uses the virtual study of bioactive molecules and the design of drug candidates that have attributes similar to those of known bioactive compounds. These technologies are commonly known as molecular modeling and computational chemistry.

Molecular modeling uses sophisticated computer programs that determine the structures and properties of molecules of interest and then intelligently analyze the data to predict the structure of an ideal drug candidate. This is no simple feat because the data that characterize molecules can exist in many formats. Carrying out that type of analysis demands extremely fast and powerful computers. Accelrys and HyperCube, among other firms, have developed software programs for molecular modeling.

Computational chemistry has also changed drug discovery. Chemists can now spend as much time at the computer as the lab bench. They use the PC to explore possible molecular configurations for their potential drug candidates. The use of programs to predict and design molecules saves the time and expense of actually screening a huge library of compounds for activity against a target. Accelrys, CAChe, and Tripos have developed computer programs to help chemists design synthetic molecules likely to have the desired biological properties while minimizing the risks of adverse effects such as toxicity.

IDENTIFYING DRUG TARGETS

Biotechnology firm CuraGen Corporation illustrates the potential value of an informatics based approach in drug discovery. It has established programs to identify and validate novel drug targets based on knowledge of the human genome. "We have focused on understanding cell biology, the gene, how genes code for proteins, and how proteins interact with other proteins," says John Murphy, vice president and chief information officer. "With that understanding we can better engineer specific drugs and therapies that can be used for pinpoint cures rather than the historical broad approach to healing."

The company identifies novel, pharmaceutically relevant genes and proteins and associates them with specific diseases using biological methods that include hypothesis-driven disease models, drug response models, gene and pathway mining approaches, and human genetics. Once associated with diseases, the genes and proteins are validated as targets through cellular assays and animal model systems.

To manage the information that results from these efforts, scientists use CuraGen's GeneScape operating portal. This consists of bioinformatics tools and databases that have been designed specifically to manage, organize, and analyze vast amounts of biological information. "The portal permits us to tie the entire world, rather than a single journal at a time, to the scientist," says Becky Horton, CuraGen's marketing manager. "We use both public and private databases. Scientists now have at their fingertips a much better platform to do their research, get references, and collect the best information on what they're trying to study."

The company has signed collaborative agreements to apply its technology. It has established strategic alliances with Abgenix to develop antibody therapeutics to treat an array of diseases and with Bayer AG to develop and commercialize small molecule therapeutics for treating obesity and diabetes.

In the past few years, biologists in both the academic and business worlds have generated huge archives in their hunt for the fundamental causes of disease. Because hundreds of labs around the world, each with different computer systems and ways of recording and storing the data, generate that information, it often comes across like the garbled e-mail message: in a form that scientific teams other than the one which produced it find difficult or impossible to understand. Plainly, continued progress of modern drug discovery at its current breakneck speed will depend on standardization of data and the format in which it is stored.

The I3C, with its plans to create a standardized global language for the biotechnology and pharmaceutical industries, represents one avenue toward that goal. All new software would follow the rules and old databases would be updated to the new specifications. Nearly 35 academic, corporate, and government organizations have taken part in discussions about the project. IBM, Sun Microsystems, and the National Cancer Institute are among the most prominent.

The BIND network database under development by blueprint WORLDWIDE takes a narrower approach to standardizing data. Based on a model created by Christopher Hogue of the University of Toronto's Samuel Lunenfeld Research Institute, it is a "living database" of bioinformatic and biomedical data intended to help the global science community move toward a complete description of the ways in which molecules interact and control cellular life. "The database currently exists at www.bind.ca and is usable with 6,000 records," says blueprint's Ouelette. "That's just a small fraction of the tens, if not hundreds, of thousands of records we hope to have in the future." The organization is filling the database with information in publicly available databases and data submitted directly by scientists.

THE EUROPEAN EXAMPLE

The research directorate-general of the European Commission (EC) has started its own efforts to help standardize databases relevant to life science. "Virtually every euro spent through our Framework program in bioinformatics goes to support standardization and linking data," says Carlos Martinez-Riera, principal scientific officer at the EC's directorate for health research headed by Manuel Hallen. Plans include the development and connection, through a new software layer, of a macromolecular database; a protein-protein interaction database; and a microarray gene expression database, a € 19.4 million ($17 million) project led by the European Bioinformatics Institute in the U.K. In addition, says Martinez-Riera's colleague Philippe Cupers, "We have a project coordinated by Per Roland of Sweden's Karolinska Institute that proposes a database generator for 3-D images of the brain, working on raw data from PET and FMRI scanners all over Europe."

Another EC group reverses the usual relationship between IT and life science. "We're looking to see what innovations we can introduce to IT from life science," says Simon Bensasson, head of the EC's future and emerging technologies unit. "We have launched initiatives on neuroinformatics — artifacts that live and grow — and autonomous sensing robots. These projects involve biologists, computer scientists, and engineers."

In the more conventional direction, computer hardware firms and software companies are working with biological researchers and with each other to provide the best possible IT support for the life science industry. Biologists worldwide can expect to experience significant advances and powerful new IT tools in the near future.

Peter Gwynne is a freelance science writer based on Cape Cod, Massachusetts, U.S.A. Gary Heebner is a marketing consultant serving the scientific industry, based in Foristell, Missouri, U.S.A.

WEBLINKS ADVERTISERS InforMax, Inc. [USA] Bioinformatics software solutions for the analysis and interpretation of genomic, proteomic and other biomolecular data. 240-747-4000 www.informaxinc.com InforMax, Inc. [UK] +44 (0) 1865 784 580 ISI Thomson Scientific [USA] Web-based and printed information providing coverage of the scientific publications and other relevant research. 215-386-0100 www.isinet.com ISI Thomson Scientific [UK] +44 (0) 1895 270 016 Messe München GmbH [Germany] Services for exhibitors and organizes trade show and scientific conferences +49 89 9 492 07 20 www.messe-muenchen.de FEATURED COMPANIES and ORGANIZATIONS Abgenix, Inc. therapeutic antibodies www.abgenix.com Accelrys molecular modeling software www.accelrys.com Amersham Biosciences AB instruments and reagents www.amershambiosciences.com Apple Computer, Inc. computers/informatics www.apple.com Applied Biosystems instruments and reagents www.appliedbiosystems.com Axaron Bioscience AG bioinformatics www.axaron.com BASF AG [Germany] chemicals www.basf.com Bayer AG pharmaceuticals www.bayer.com Biomax Informatics AG [Germany] bioinformatics www.biomax.de Bio-Rad Laboratories — Life Science Group instruments and reagents www.discover.bio-rad.com Biotechnology Industry Organization (BIO) industry support organization www.bio.org blueprint Worldwide, Inc. bioinformatics www.blueprint.org CambridgeSoft Corporation cheminformatics www.camsoft.com Celera Genomics bioinformatics www.celera.com Centre for Molecular Medicine and Therapeutics Bioinformatics Core Facility research organization www.cmmt.ubc.ca Compaq Computer Corporation computers/informatics www.compaq.com Compugen, Inc. bioinformatics www.cgen.com Cray, Inc. supercomputers/informatics www.cray.com CuraGen Corporation biopharmaceuticals www.curagen.com Entigen Corporation bioinformatics www.entigen.com European Bioinformatics Institute (EMBL) bioinformatics www.ebi.ac.uk Fujitsu America, Inc./CAChe Group scientific software www.cachesoftware.com GeneBio SA bioinformatics www.genebio.com GeneProt, Inc. scientific software www.geneprot.com Genomatix Software GmbH scientific software www.genomatix.gsf.de GraphPad Software, Inc. scientific software www.graphpad.com HyperCube, Inc. molecular modeling software www.hyper.com IBM Life Sciences computers/informatics www.ibm.com InforMax, Inc. bioinformatics www.informaxinc.com Interoperable Informatics Infrastructure Consortium (I3C) industry support organization www.i3c.org Karolinska Institute university/research organization www.ki.se Lawrence Livermore National Laboratory government research institute www.llnl.gov LION Bioscience AG [Germany] bioinformatics www.lionbioscience.com Lynx Therapeutics, Inc. genomics kits and systems www.lynxgen.com MDL Information Systems cheminformatics www.mdli.com MDS Proteomics proteomics/bioinformatics www.mdsproteomics.com Micromass UK, Ltd. mass spectrometers www.micromass.co.uk Motorola microprocessors www.motorola.com National Cancer Institute government research institute www.nci.nih.gov National Center for Biotechnology Information (NCBI) government research institute www.ncbi.nlm.nih.gov Oxford GlycoSciences (UK), Ltd. proteomics/bioinformatics www.ogs.com SAS Institute, Inc. scientific software www.sas.com Sigma-Aldrich Corporation kits and reagents for research www.sigma-aldrich.com Sun Microsystems operating systems/informatics www.sun.com TimeLogic Corporation bioinformatics www.timelogic.com Tripos, Inc. cheminformatics software www.tripos.com TurboGenomics, Inc. scientific software www.turbogenomics.com University of Toronto's Samuel Lunenfeld Research Institute research organization www.msri.on.ca

Note: Readers can find out more about the companies and organizations listed by accessing their sites on the World Wide Web (WWW). If the listed organization does not have a site on the WWW or if it is under construction, we have substituted its main telephone number. Every effort has been made to ensure the accuracy of this information. The companies and organizations in this article were selected at random. Their inclusion in this article does not indicate endorsement by either AAAS or Science nor is it meant to imply that their products or services are superior to those of other companies.

This article was published as a special advertising section in the 8 March issue of Science