Where worlds collide: the University of California, San Diego, is making a concerted effort to bridge the gaps between biology and physics.
Fresh approaches to complex biological systems and the need to organize and make use of the mountains of data coming out of genomics have created an unprecedented demand for scientists who can straddle the interface between physics and the life sciences.
Physicists and biologists will have to work alongside each other to clear this century's highest scientific hurdles, even though cultural differences and old-fashioned attitudes can cause friction. To accelerate collaboration and train the new interdisciplinarians, many universities are bringing theoreticians and experimentalists together. Large numbers of interdisciplinary departments and graduate programmes are now starting up at universities around the world.
At the Center for Theoretical Biological Physics at the University of California, San Diego, co-director José Onuchic is delighted with the success of the centre's precursor programme, La Jolla Interfaces in Science, which trains postdocs and graduate students to apply quantitative methods to biology. "We were told: 'the students won't be real physicists, chemists or biologists and they're not going to get any jobs'," he says. "But it was exactly the opposite."
Nearly two-thirds of the postdocs on the programme have gone on to an assistant professorship, says Onuchic. And many of the rest have moved into high-paying positions in industry working in drug discovery or materials science.
José Onuchic (right) and his team work at the interface between theory and experiment.
With US$10.5 million in funding spread over five years from the National Science Foundation, the centre fosters the development of physical approaches to biological systems and new forms of bio-inspired physics. It was set up with the goal of creating a two-way exchange between theoretical and molecular biology groups. To achieve this, each graduate student or postdoc is attached to two laboratories in different disciplines.
This approach is catching on — Harvard University's Bauer Center for Genomics Research, which opened last year, takes a similar tack. It is an angle that attracted Aviv Regev, a computational biologist who recently took up a fellowship at the centre. She strongly agrees that studying both benchtop biology and computational science contributed to her own success. Theoretical students need to learn first-hand how experimental biologists solve problems, she adds.
"Your ongoing science is going to be composed of moving between two very different systems," says Regev, who says she constantly switches between thinking intuitively about a biological problem and fashioning it and a possible solution into a computational problem with known parameters.
It is not only life scientists who are seeking new partnerships to help them to tackle complexity; physical scientists also increasingly seem to be looking for ways to branch out.
This trend is reflected in recent meetings. In September 2002, the American Physical Society held a meeting entitled 'Opportunities in Biology for Physicists' that attracted 200 young physicists. In Britain, the Engineering and Physical Sciences Research Council created the Life Sciences Interface in 1999. The programme funds large group grants of £10 million (US$16 million) for medical imaging, bio-nanotechnology and tissue engineering, and facilitates cross-talk workshops and informal week-long meetings that throw biologists and physicists together to see if they can come up with new ways to solve biological puzzles.
"One of the things most often given as a barrier to working at the interface is time," explains John Hand, the programme's director. "So we created the Discipline Hopping scheme." This offers individual researchers up to £50,000 to venture into another field by taking a sabbatical, travelling or hiring a cross-discipline research assistant for the lab.
Scientists who can move fluently between disciplines may end up with double the academic opportunities: biology departments are searching for bioinformatics specialists and computational biologists, whereas physics departments are looking to expand into biological physics.
COMPUTING POWER
If any one group within biophysics is set to cash in on the genome, it is computational biologists. Demand for them is skyrocketing as the number of sequenced genomes rises. According to most researchers in genomics, a severe shortage of computational biologists remains the biggest bottleneck in the task of turning sequenced genomes into meaningful interpretations.
"Unfortunately, there are not enough people who have both an understanding of biology experiments and of the algorithms involved in bioinformatics," says Lakshmanan Iyer, a computational biologist at Harvard who has provided technical support for bioinformatics research at both Harvard Medical School and the Bauer Center for Genomics Research. Iyer trains scientists to use web-based sequence-analysis tools and also advises on individual project strategies.
"If you have a specialization in statistics or maths with exposure to biology, then you are really set," he says. The International Society for Computational Biology's website job list reflects the wealth of opportunities, with almost 40 academic, research institute or genetic-based biotechnology positions posted in the past three months.
Although many computational biologists are being snatched up to work on the genome level, others are concentrating on the next level of integration, the pathway. Systems biology has got off to a flying start as the research community has come to realize that understanding the 'interactome' — all of the possible protein–protein interactions for a cell — will be the next big hurdle in biology.
Aviv Regev believes that a flexible approach is essential for computational biology
To prepare for the challenge, systems biologists are beginning to form coalitions and consortia — much like in the early days of the Human Genome Project. The Alliance for Cellular Signaling has a US$25-million National Institute of General Medical Sciences (NIGMS) grant as well as corporate sponsorship. It plans to track all signalling events in mouse immune and heart-muscle cells.
Other efforts at integration include BioPAX, a public database for keeping track of known pathway information, and a new computer modelling language, Systems Biology Markup Language, or SBML, that will allow researchers directly to compare computational models of biological systems.
Regev looks forward to the day when her models can be matched up easily with others. She is applying new forms of calculus, originally designed for mobile-phone networks, to cellular signal-transduction cascades. The phone networks, she says, are analogous to biological systems, where molecular interactions happen concurrently rather than linearly and determine "who speaks to whom". She is betting that quantitative modelling of well-known pathways such as the MAP kinase cascade will yield better predictions of interactions.
STRUCTURAL BOOM
But before jumping from completed sequence to complete systems, there is a big bridge to be built, and with significant labour from biophysicists: solving the 'protein structure genome'.
Nuclear-magnetic-resonance spectrophotometers, such as this 800-MHz machine, are key to elucidating protein structures.
This is the field of structural genomics — the marriage of genomics, computation and experiment that aims to speed up the determination of structures of most of the proteins whose structure and function are still unknown. Ambitious structural-genomics initiatives are under way in the United States, Germany and Japan, in particular, and promise to increase the workload at synchrotron X-ray and nuclear-magnetic-resonance (NMR) facilities. The European Union's Sixth Framework Programme has designated 
1.1 billion (US$1.2 billion) for bioinformatics and structural-genomics work from 2002 to 2006.
"We need to solve 20,000 structures to characterize about 90% of the human genome," says Andrzej Joachimiak, director of the Structural Biology Center at Argonne National Laboratory's Advanced Photon Source (APS) in Illinois. He and others say that it will take the concerted effort of many scientists trained in structural biology, including those with skills in protein purification and crystallization, to complete this task. Joachimiak is also principal investigator for the Midwest Center for Structural Genomics, one of nine structural-genomics consortia supported by the Protein Structure Initiative, a NIGMS programme.
With a projected half-a-billion dollars in funding, the initiative seeks to add 10,000 new unique structures to the Protein Data Bank in ten years. Ideally, that would result in one representative protein structure for each protein family, which would serve as templates for solving other members of the families.
"It could create a catalogue for a short cut to all of the protein structures on the planet," says Charles Edmonds, a programme director at the NIGMS. As such, the nine centres are not limited to human protein structures; their subjects range from plants and nematodes to the bacteria that cause tuberculosis.
But to accomplish this feat, each centre must significantly ramp up its production in the next two years to churn out 200–300 structures each year. Last year Joachimiak's centre solved 45. To boost output, the NIGMS, together with the National Cancer Institute, has agreed to build three new synchrotron X-ray beamlines at the APS at a total cost of about $20 million.
The beamlines provide high-brilliance X rays that can resolve large and complex molecular structures in a fraction of the time needed with conventional X-ray sources. For example, Joachimiak and his colleagues solved the structure of a small bacterial protein in under 7 hours. More typically, he says, you can go from crystal to structure in about 30 hours for proteins that might have taken three years with earlier methods.
The advent of structural-genomics initiatives has raised the demand for beamtime at existing facilities, prompting the construction of new synchrotrons and additional beamlines at existing ones. Gerhard Materlik, chief executive of Diamond, the new UK synchrotron scheduled to open in early 2007 in Chilton, says that about half of the first-phase beamlines will be dedicated to life-science research. He anticipates hiring 200–250 technical staff by 2010.
"We're really looking for qualified experts," he says. "There is not an abundance of people who can run these machines." Competition will be fierce — other projects scheduled to come online include the Synchrotron Light Laboratory in Barcelona, the Australian Synchrotron in Melbourne, and the Canadian Light Source in Saskatoon. These facilities require physicists to build, develop and maintain the beamlines, crystallographers to advise users and help to collect data, and administrators to make decisions that will guarantee efficiency. Even well-funded facilities have trouble keeping up with demand and finding enough staff to run experiments around the clock.
"I compare these facilities to driving a very sporty car, like a Ferrari, 24 hours a day, seven days a week," says Joachimiak. Running state-of-the-art technology can mean fast-paced excitement and constant problem-solving challenges, but it can also easily lead to burnout when helping hundreds of users each year.
Managers agree that these positions, although comparable in salary to an assistant-professor position, can be frustrating for someone who wants to develop their own research programme. Most of the job is devoted to helping users run experiments and trouble-shooting. To keep highly trained technical staff stimulated, some facilities are encouraging ongoing research collaborations for their workers (see 'A joint effort'). And, Joachimiak points out, unlike an academic position, facility funding is extremely stable and doesn't require grant writing.
The cheaper and faster structure determination becomes, the more attractive it is to pharmaceutical companies. Ten major companies share a beamline at the APS for high-throughput screening of hundreds of drug candidates bound to their target protein. And the drug-discovery company Structural GenomiX, based in La Jolla, California, built its own beamline at the APS for a gamble on structure-based drug discovery.
A German initiative — the Protein Structure Factory — hopes to crank out human protein structures based on the German human genome project. This high-throughput assembly line starts with complementary DNA sequences and ends with crystal-structure determination done at the BESSY II synchrotron in Berlin. These large-scale endeavours will need biology-literate computer scientists who can optimize the beamline system for automation.
MAGNETIC MOMENTS
Online: drug-discovery firm Structural GenomiX has its own X-ray beamline at the Advanced Photon Source in Argonne.
Another technological development that will make an important contribution to the structural-genomics initiative is the use of high-field NMR spectrophotometers. These machines use a powerful 900-MHz magnet to reduce analysis time and boost resolution. They can be used to tease apart large protein complexes or to examine protein–protein and protein–DNA interactions. "It brings experiments into reality that would have otherwise been shelved," says Paul Ellis, director of the macromolecular structure and dynamics group at the Pacific Northwest National Laboratory in Richland, Washington.
But such facilities are still relatively rare — there are just two in the United States, at Pacific Northwest and the Scripps Research Institute in La Jolla. This should soon change, as the National Institutes of Health has recently awarded grants for four machines to the Massachusetts Institute of Technology in Cambridge, the New York Structural Biology Center in New York City, the University of Georgia in Athens, and the University of Wisconsin, Madison.
Mastering technology outside a given researcher's discipline is an important step towards broader career options. Anthony Watts, who directs the National Biological Solid State NMR Facility at the University of Oxford, UK, encourages flexibility among his staff. "I try to expose all of the people to as many technologies as possible," he says. "If they want to be independent in academia or industry, they need to take on as many tasks as possible."
Watts' lab has an 800-MHz NMR spectrophotometer, which he uses to study membrane proteins that might serve as drug receptors. He says that physicists in his lab learn how to culture bacteria, whereas the biologists find out how to design NMR experiments.
Demand within the marketplace for those who master the range of technical skills is strong. "Technical staff is a seller's market," says Edmonds. "There are just not enough people who have a spectrum of skills that spans a detailed understanding of computer-controlled instrumentation and the details of experimental structural biology."
This is especially good news for anyone with dual training or who wants to move from physics or biology into biophysics. A survey by the US Biophysical Society of biophysics students and postdocs in 1999 showed that 84% believed they would land a position in their chosen field — incredible optimism at a time when the National Research Council reported that, for the life sciences in general, there were too many people being trained.
Their ability to sift through stacks of data quantitatively and arrive at biologically relevant answers looks set to provide biophysicists with a secure future in the biology lab. "There is an enormous amount of data that has to be turned into a picture — there are a lot of opportunities out there," says Iyer.
Web links
Center for Theoretical Biological Physics
http://ctbp.ucsd.edu
Bauer Center for Genomics Research http://cgr.harvard.edu
Advanced Photon Source http://www.aps.anl.gov
Pacific Northwest National Laboratory http://www.pnl.gov
Protein Structure Factory http://www.proteinstrukturfabrik.de
Diamond http://www.diamond.ac.uk
