The Rise of Systems Biology

First published August 2004

"Biology is making an historic transition from being a descriptive science to being an engineering science."

Nanotechnology is so yesterday.

The nanotechnology future, of microscopic machines made of silicon gears and motors, may soon look as dated as Disney's Tomorrowland. Imagine instead a future in which living machines -- bacterial robots, if you will -- travel through the human body cleaning up cardiovascular plaque and making repairs. Or latching onto cancer cells and releasing drugs directly into tumors, all the while emitting a dye that enables imaging machines to track their progress. This would be machine-making at its natural and fantastic best. Yet is it fantasy?

Many scientists practicing what they call "systems biology" don't think so. For them, being able to understand how all the pieces of a biological system interact, rather than understanding each piece individually, allows the construction of completely new organisms. It is also the logical consequence of more than two decades of research advances and vastly superior computing power. What once seemed impossibly complex has now become merely complicated. And while engineers may not consider systems biology a particularly novel concept, its advocacy represents a qualitative change in the way biologists approach the most important scientific questions of the day.

"Biology is making an historic transition from being a descriptive science to being an engineering science," says Regis Kelly, director of the Institute for Quantitative Biomedical Research (QB3). "This transition is putting us on the edge of one of the most exciting times in the history of biological science, which is why systems biology is such a hot field right now."

In order to practice biological engineering, the scientists say, they have to understand organisms in a different way.

"In systems biology we try to describe living systems completely, down to the molecular level," says Ajay Jain, director of the UCSF Comprehensive Cancer Center's bioinformatics core. "The ultimate goal is to describe all aspects of the system computationally. We are obviously a long way from doing this in complex organisms, but we are getting closer in simple organisms."

Old vs. New Biology

The new laboratories at UCSF might look entirely different from those that researchers inhabited in the institution's early days a century ago. Glassware and Bunsen burners have been supplemented by computers, gene-sequencing machines, high-speed centrifuges and mass spectrometers. Yet until recently, research methods had changed very little. Standard practice has been to isolate a molecule, gene or protein and perform in-depth analysis of what it does and how it works. Whole careers have been built on one gene or one hormone.

Systems biology introduces a new approach in which scientists are more interested in how the pieces of the puzzle fit together, how the molecular players interact, than in a detailed analysis of each puzzle piece.

The difference between these two methods was highlighted by Cold Spring Harbor scientist Yury Lazebnik in a satirical article entitled "Can a Biologist Fix a Radio?" in the journal Cancer Cell. A biologist, Lazebnik wrote, might shoot metal objects at many similar radios and then take apart the radios that stopped working after the onslaught. If the biologist finds a broken part in a radio that does cease functioning, he might name this part the Most Important Component (Mic). If another part also stops the radio from working, he might name it Really Important Component (Ric). Describing the color, structure and electrical properties of Mic and Ric might ensure decades of grant money and a distinguished career, but get him little closer to understanding how a radio works.

An engineer, on the other hand, would want to construct a wiring diagram showing how all the pieces of the radio connect with each other. Even if he doesn't understand at first how each part works, he could make testable guesses based on how electrons have to flow through the parts to transform a radio signal into sound.

Another metaphor that UCSF School of Pharmacy researcher Ken Dill uses to describe the differences between classical biology and systems biology is the Boeing 777 aircraft. "You could study every part and describe how each functions, but that still wouldn't tell you how the airplane flies," Dill says.

An Addiction to Data

The Boeing 777 metaphor is a good one because that airplane was the first to be designed and initially tested wholly on computers before a single part was built. This effort demanded a vast quantity of data and powerful computer models, which is also true for systems biology. Quantitative models of a single cell may be built on millions of data points and require computers to handle them.

Of course the field of systems biology doesn't just need computers and other data-rich technologies. It is also the product of the new technologies. It came into being partly out of a need to handle the large volumes of data produced by new techniques. Automated gene sequencing technology has produced complete gene sequences for over 100 organisms. At the same time, microarray technology allows scientist to quickly assay the activity of thousands of genes or proteins simultaneously. In general, this has been a great thing for biology.

"The neat thing is that the cost per measurement of each data point has become extremely small," Jain says.

But bushels of cheap data points also present scientists with more information than they can digest using the classical model of biology, in which each gene or protein is studied individually. The results can't be bitten off piece by piece; they have to be digested whole.

"Everything changed in the last five years," UCSF researcher Chris Voigt says. "We are now able to perturb a cell and monitor how it responds on a large scale of measures."

Such a systematic approach has been a huge boon for cancer researchers. UCSF scientists J. Michael Bishop and Harold Varmus' discovery that normal cellular genes can be converted to cancer genes, called oncogenes, won them the Nobel prize and was a huge leap forward. But it turns out that oncogenes are usually just one part of a cascade of gene changes that can turn a good cell bad.

"Cancer is much more a problem of the cell system than a problem with a single gene," says UCSF researcher Donna Albertson, who is using microarray technologies combined with comparative genomic hybridization to identify the multiplicity of genes that have changed their usual activity in cancer cells. "To become cancerous, a cell has to change in many ways, overcoming numerous safeguards that normally keep renegade cells in check."

Investigator Joe Gray, who has a joint appointment at UCSF and Lawrence Berkeley National Laboratory, is using similar technologies to pick up genes that are commonly deregulated in breast and ovarian cancer. He and others are using this information to identify new therapeutic targets and to develop strategies to predict individual responses to therapeutic drugs. "These techniques provide a more comprehensive view of the genes that are involved in cancer progression. We can use this information now to improve cancer classification. Eventually, we expect it will lead to a much more accurate understanding of the genetic and biological processes that are involved in cancer," Gray says.

Page 2 - Building Models

Regis Kelly
Regis Kelly, Director of the Institute for Quantitative Biomedical Research. Photo by Majed Abolfazli.
Ken Dill
Ken Dill, Pharmaceutical Chemist, UCSF School of Pharmacy. Photo by Majed Abolfazli.
Ajay Jain
Ajay Jain, Director of the UCSF Comprehensive Cancer Center, Bioinformatics Core. Photo by Majed Abolfazli.
Donna Albertson
Donna Albertson, Researcher, UCSF Comprehensive Cancer Center. Photo by Majed Abolfazli.
Joe Gray
Joe Gray, Program Leader and Researcher, UCSF Comprehensive Cancer Center. Photo by Majed Abolfazli.

Related Links: