Home | This Week's Contents  |  C&EN ClassifiedsSearch C&EN Online

 
Millennium Special Report
C&EN 75th Anniversary Issue
 
Employment Feature Advertisers

AstraZeneca

Baxter Healthcare corp.

ChemoCentryx, Inc.

Millennium Pharmaceuticals

Novartis

Xencor

Related Stories
FINDING A HOME IN THE BIOSCIENCES
[C&EN, September 17, 2001
]

BIOINFORMATICS
[C&EN, February 19, 2001
]

THE HOTTEST JOB IN TOWN
[C&EN, January 1, 2001]

Career & Employment Archive

E-mail this article to a friend
Print this article
E-mail the editor
 
 
 
 
 
 
 Table of Contents
 C&EN Classifieds
 News of the Week
 Cover Story
 Editor's Page
 Business
 Government & Policy
 Science/Technology
 Concentrates
  Business
  Government & Policy
  Science/Technology
 Education
 ACS News
 Calendars
 Books
 Digital Briefs
 ACS Comments
 Career & Employment
 Special Reports
 Letters
 Newscripts
 Nanotechnology
 What's That Stuff?
 Pharmaceutical Century

 Hot Articles
 Safety  Letters
 Chemcyclopedia

 Back Issues

 How to Subscribe
 Subscription Changes
 About C&EN
 Copyright Permission
 E-mail webmaster
EMPLOYMENT
January 7, 2002
Volume 80, Number 1
CENEAR 80 01 pp. 47-53
ISSN 0009-2347
[Previous Story] [Next Story]

CAREERS IN BIOINFORMATICS
Field is not significantly affected by economic downturn; qualified people are still hard to find

CELIA M. HENRY, C&EN WASHINGTON

With the ever-expanding reach of genomics and proteomics and the large amounts of data that their associated techniques generate, biology is becoming an increasingly quantitative science. The need is growing for scientists--whether they be trained as molecular biologists, biochemists, or chemists--who can speak the languages of both the computational and biological sciences.

8001new_about

COMPUTER WHIZ Bioinformaticists must be comfortable with both biology and computer science.
ACCELRYS PHOTO

The field of bioinformatics is one way to deal with this explosion of data. Unfortunately for neophytes trying to understand the field, bioinformatics means different things to different people. At its core, however, the field is simply the intersection of biology and the computational sciences, which include mathematics, statistics, and computer science.

Nathan O. Siemers, associate director of applied genomics at Bristol-Myers Squibb in Hopewell, N.J., calls bioinformatics "a mixture of the mundane and the sublime." The mundane includes data integration, data formatting and conversion (also known as parsing), and automation--"doing a relatively straightforward analysis a million times over the course of a few days and then dealing with the volume of results you get and turning them into something useful," Siemers says.

It's when the useful parts start to emerge that the sublime aspects surface. "Some of the most exciting stuff is working with large bodies of sequence and gene-expression data," he says, "seeing the patterns that come out and how they reproduce biology, in some cases extremely well."

FOR MANY PEOPLE, bioinformatics is focused on analyzing genomic data, but for others it's much broader. Originally confined to the early stages of drug discovery, bioinformatics is lending its power to all aspects of clinical development, a trend that will only continue. "In industry, I see bioinformatics evolving from primarily supporting early discovery to becoming more embedded in all stages of the drug discovery process," Siemers says. "We're already established in pharmacogenomics, trying to tie efficacy and side effects in a clinical setting back to underlying genetic variation in our patients."

Much bioinformatics work at Bristol-Myers currently focuses on RNA measurements, such as gene-expression profiling, Siemers says. However, proteomics is catching up. "It will probably take another couple of years before the volumes of data from those [proteomics] efforts really equal that of the RNA profiling chips," he says.

Abgenix, located in Fremont, Calif., is also taking a broader view of bioinformatics. Its informatics group is subdivided into a software development group, a computational biology group, and even a library information science group. "We're involved in the same types of genomics data analysis that you have in a typical bioinformatics group, but we're also involved in lab automation and algorithm development for the various high-throughput assays, preclinical assays, and process development," says Keith Joho, director of bioinformatics. "We're putting together an integrated bioinformatics approach that not only encompasses that very beginning part--the genomics data analysis--but really covers the spectrum of drug discovery up to going into the clinic."

Rosetta Inpharmatics, a subsidiary of Merck located in Kirkland, Wash., takes a quantitative and computational biology approach to bioinformatics, according to Eric Schadt, a chief scientist and head of the computational genomics program.

At Rosetta, bioinformatics falls under the larger umbrella of informatics, headed by Roland Stoughton, and is divided into three groups. A custom data analysis group headed by Hongyue Dai and Yudong D. He develops solutions to process large sets of RNA expression array data and integrates these data with other sources of information. These individuals must take the data from thousands of arrays--millions of measurements--and "recognize the patterns within those millions of measurements that really elucidate the processes of interest in the system under study," Schadt says. The individuals in that group tend to have doctorates in mathematics or statistics.

The second group is the computational genomics program, which Schadt describes as being closer to classic bioinformatics. That group does sequence-based analyses, such as annotating genes and proteins, searching for motifs in proteins, or predicting genes in genomic sequence. However, that group is also "directly involved in driving the biology and thinking of innovative ways to integrate expression, genetics, genomic, and other data of interest for the purposes of discovery." Most of the people in that group have a Ph.D. in math, statistics, or biology. Schadt himself has a Ph.D. in biomathematics. The third group, headed by Mihai Margarint, focuses on software engineering and provides advanced software development support to the other two research teams.

ALTHOUGH BIOLOGISTS first come to mind when talking about bioinformatics, chemists are also welcome. "Chemists who are more biologically oriented, like biochemists, play a big role in helping us piece together pathways and understand protein function," Schadt says. "Also, their training, especially in areas like physical chemistry, tends to be more quantitative. They're more adept at the sort of analyses we do than a classically trained biologist, who typically doesn't get the math and statistical training that chemists often get."

In fact, the small number of biologists working within the informatics team at Rosetta "tend to be those biologists with strong programming, database, or quantitative skills, which are skills many graduate programs in biology haven't integrated into their curricula," Schadt says. In contrast, physical chemists and physicists tend to be more skilled in developing and applying numerical algorithms to mine data. "I think the need for people who are trained more quantitatively just grows and grows as more people in the biological sciences accept that biology is becoming more of a quantitative science. The quantitative training will continue to enable research in this field to be more competitive," he says.

At Large Scale Biology Corp. in Vacaville, Calif., Gary M. Wolfe, vice president of informatics systems, prefers to pull bioinformaticists from the physical sciences because they tend to have a "good mathematical foundation." They may be required to develop user interfaces for existing algorithms or to develop algorithms of their own. "I'd rather get individuals who are heavy-duty experienced scientists with a strong programming background," Wolfe says. "Then it's pretty straightforward to write a Web interface to the BLAST [sequence alignment] algorithm. But that individual on another day could put together a sophisticated algorithm to analyze mass spec data."

Wolfe believes strongly that a bioinformatics platform must effectively integrate laboratory information management systems with data analysis and discovery applications. The data generated must be stored and tracked in a relational database. "Once the data are in some ordered format, it becomes straightforward to write programs for analyzing the data, and generally speaking, this enables us to create a discovery platform," Wolfe says. "However, to create such a platform, it is necessary to have individuals with excellent systems analysis and programming skills."

However, Wolfe says, it's not possible to take just any person with a programming background or mathematical background and have them program for bioinformatics applications. "They have to have an interest and experience in the life sciences. That can come from previous work they've done in a company or even their graduate or undergraduate training."

"The need for people who are trained more quantitatively grows as more people in the biological sciences accept that biology is becoming more of a quantitative science."
COMPUTER SKILLS are very important for work in bioinformatics, especially UNIX and the programming languages C and PERL, Wolfe says. He focuses on these particular languages because most bioinformatics programs are written in C and because PERL is powerful for text processing and is flexible with Web applications.

Schadt agrees that ideal bioinformaticists will be strongly grounded in math, statistics, and computer science. In addition, they will have at least some level of biological understanding. "The ones who are the most productive are the ones who can set up databases themselves, write scripts and programs themselves to process data or assemble data in a meaningful way, and then statistically summarize those data for the purposes of biological discovery. As the bar gets raised, I think a better understanding of biology, having the ability to identify the really pressing questions, is going to be required to be really competitive."

Joho says the ideal preparation varies because bioinformatics requires a variety of people. "The key would be that you have to be very good in one area--for instance, an accomplished biologist or biomathematician--but you should also have some additional training in another area. For example, a mathematician who has a good biology background is very valuable, as is a biologist who has a computer science or mathematics background."

At Consensus Pharmaceuticals in Medford, Mass., the best candidates will have strong biology backgrounds, "because the computer skills are broadly available in a variety of candidates," says John L. Krstenansky, vice president of chemistry. "You have to know how to interpret the data and know what's really meaningful if you're generating algorithms."

Krstenansky considers people with purely computer science backgrounds or who have been out of the science for a while not to have "the scientific strength" needed to build a strong bioinformatics program. "You need a strong chemistry or biology background that cannot be achieved without formal training. You must be able to be recognized as a contributing scientist in that area of science. That's the kind of depth of understanding and currency in the field that I think is essential."

The bioinformatics needs at Consensus lie outside the "mainstream bioinformatics" of genomic data, Krstenansky says. The company, which is based on a library-screening technology for identifying ligands to various enzymes or receptors, needs people who can assess information to identify "new biology." Therefore, Consensus is particularly interested in candidates with a strong background in biochemistry and an understanding of cellular biology.

Although Krstenansky anticipates that the bioinformatics effort at Consensus will grow, he doesn't see it reaching the size of the company's biology and chemistry groups. "Your primary need is people generating the real materials and the raw data to be analyzed," he says. "If things are being done right, you should need fewer people proportionately to do the analysis of the data or to create the tools for the analysis of the data."

8001wolfe3

FUNDAMENTALS Wolfe believes that strong training in math, chemistry, and physics creates a strong foundation on which to build a bioinformatics career.

SOME COMPANIES focus on providing bioinformatics software to other companies. One such company is Accelrys, located in San Diego. The "ideal profile" for someone in bioinformatics at Accelrys is someone who has at least a master's degree in biology in addition to a computer science degree, according to Judith M. Ohrn, vice president of human resources. "That's really nirvana," she says. If they can't find someone with training in both biology and computer science, they will hire a strong computer scientist who can pick up the biology on the job. "Normally you find with scientists who are good engineers or good computer scientists a certain fundamental knowledge about biology and an aptitude to learn what they need to learn."

Universities are grappling with the issue of what to include in a bioinformatics program. C. Fred Fox, a professor of microbiology and molecular genetics who is actively involved with the bioinformatics program at the University of California, Los Angeles, asks: "What training must these people have so they can sustain a career without going stale?" At UCLA, students must fulfill the requirements for a degree in one of the 13 participating departments--including math, statistics, and biochemistry--and then take additional courses to earn the bioinformatics certificate. Programs are being set up at both the master's and Ph.D. levels. The Ph.D. certificate program has already been training students for three years and should soon receive its final approval from the faculty senate.

The main issue in preparing for a career in bioinformatics isn't whether an individual is a biologist or computer scientist first, says Michael N. Liebman, director of computational biology at the Abramson Family Cancer Research Institute at the University of Pennsylvania. He spent 12 years working on bioinformatics in the pharmaceutical industry. "You can come in with a molecular biology background and learn computer science. You can come in with a computer science background and learn molecular biology. The issues that really need to be addressed are how do you solve complex problems and how do you work in teams."

Liebman complains that students are not usually taught how to identify the right question about a problem. "They're taught how to apply methodologies if the question is presented to them," he says. "You don't need an algorithm to run faster. You first need to ask if it's the right algorithm and if it's solving the right problem."

Liebman calls bioinformatics a "moving target." People who are training for a career should not expect that the qualifications for the jobs advertised today will also be what is needed in the future. "You really have to look for a training program that's looking ahead to see what the longer term issues are and not just turning somebody out for an industrial position."

Opinions vary over the value of specialized bioinformatics programs. For example, Wolfe worries that such programs teach very little in each of the underlying areas. "I'm a real proponent of the fundamentals," he says. "The fundamentals are math, chemistry, and physics. With that, you've created a great foundation."

Schadt believes that graduates of master's-level programs may find their career choices limited by not having a doctorate. "If you're at all passionate about the science, it's a very tough world to be in where you're at the low end of the totem pole, really servicing the scientific efforts," he says. "It's incredibly difficult to drive projects at the scientific level without having jumped through the Ph.D. hoop."

No matter what their educational background, people need to be nimble and willing to seize fast-moving opportunities. "If you're going to be in this field, it's a really dynamic field with small windows of opportunity," Wolfe says. "If you can't keep up with the Joneses and keep moving and keep learning new things, then it's not a good place to be. That's just the baggage that goes along with the fact that this field is progressing so fast and will be around for a long time."

EVEN WITH THE downturn in the economy, the demand for bioinformaticists does not seem to have significantly diminished. "The field is somewhat insulated because there is still such a high demand for skilled bioinformaticists," Wolfe says. "The fact is that there is a lot of money going into genomics and proteomics these days; I assume that the demand for bioinformatics will only grow over the coming years."

Schadt's experience has been that several bioinformatics companies have instituted hiring freezes. However, he says, "the bioinformatics-based groups are less affected" because "special allowances can be made to hire in hotshot types." In the short term, he says, lower level bioinformaticists will be the most affected, including people with master's or Ph.D. degrees with little or no experience.

"But for any of these big companies that have a freeze on hiring, there are other big companies and smaller start-ups that are more eager than ever to pick up bioinformatics types," Schadt says. "The bottom line, I believe, is that the bioinformatics field will not feel any real pain unless the economy gets really bad."

Schadt believes that the bioinformatics field will continue to be insulated from all but major economic shocks for several reasons. It is still a relatively new field, and there are not enough qualified people to fill all the positions. Plus, companies and academic centers continue to realize the need for bioinformaticists and thus create new positions.

Even though the human genome has been sequenced, don't expect the need for bioinformaticists to vanish anytime soon. "I couldn't think of a better place to be in the next five to 10 years," Wolfe says. "Five years from now, people are going to look back at even the year 2000, give a little chuckle, and say, 'If these people had the data we're working with today, they could have done this, this, and this.' Ten years from now, they're going to be saying the same thing about the year 2005, let alone 2000."

[Previous Story] [Next Story]



Top


Chemical & Engineering News
Copyright © 2001 American Chemical Society


Home | Table of Contents | News of the Week | Cover Story
Business | Government & Policy | Science/Technology
Chemical & Engineering News
Copyright © 2001 American Chemical Society - All Right Reserved
1155 16th Street NW • Washington DC 20036 • (202) 872-4600 • (800) 227-5558


CASChemPortChemCenterPubs Page