November 1, 2002
Creating virtual life
Project CyberCell is making headway in an international effort to create life on a chip
"In 10 years this is not going to be pie in the sky." Excitement creeps into Dr. Mike Ellison's voice when he imagines what his research could lead to some day: Project CyberCell aims to create a virtual representation of a living cell, and ultimately more sophisticated organisms, which would behave digitally in the same manner as the real thing.
"This will lead to smarter, faster, cheaper science, because we can use the computer to tell us what the most appropriate experiments are to test and design drugs," said Ellison, executive director of the University of Alberta's Institute for Biomolecular Design. "There will be a day when we test chemicals not only for their intended uses but also for their side effects.
"We might even be able to make evolutionary predictions: say, for example, we want to develop an organism that could consume PCBs. We could imagine adding PCBs to the virtual environment, pressing the .evolution button' and allowing the organism to evolve genetically so that it could exploit a toxic environment. Then, all of a sudden, we've got the ability to deal with PCBs.
"We may even be able to program simple organisms as smart pills - nano-devices that target themselves to sites of disease and dispense drugs cleverly by anticipating changes in human physiology."
Ellison is willing to speculate on the outcome of Project CyberCell but is equally frank about its complexity. Our biological history is written in the three billion letters of genetic code contained in each of our one hundred trillion cells. Mapping the human genome is considered the greatest scientific achievement of our time, but Project CyberCell, and its counterparts around the world, is involved in work that will make the Human Genome Project look simple.
To create computerized life, researchers need to know everything that goes on in a cell and why. To do that, scientists need to be grounded in proteomics, the identification of all the proteins in a cell, and the understanding of how those proteins work together. The idea is that understanding how life itself works will reveal how disease works and how to best cure or prevent illness.
But mapping the human proteome is a greater challenge than mapping the genome. DNA consists of four chemical bases: adenine, cytosine, guanine and thymine. Proteins, on the other hand, are constructed from 20 different amino acids. Even when researchers determine the amino acid composition of a particular protein, other intimidating questions arise: what does the protein do? How does it interact with other proteins? The complexity of the challenge goes on: the human genome contains some 40,000 genes, each capable of creating one or more proteins. Different organs produce different types of proteins too: the pancreas makes one set; the brain makes another.
Because of that level of complexity, Project CyberCell is focusing on a smaller, more achievable goal. It aims to create a virtual version of the bacterium Escherichia coli. "It's the simplest organism that we know the most about," said Ellison. "You gotta walk before you can run."
The bacterium is relatively simple, with 4,000 genes each producing one type of protein. Ellison's goal is simple too, but not easy. "We want to find out everything that happens when conditions change." For example, when a cell's temperature changes, Ellison and his team want to know which proteins increase and which decrease; they want to know the rates at which a gene synthesizes a protein, the rate at which the protein carries out its function and the rate at which it degrades. "We need reliable numbers to drive and validate our simulation."
The CyberCell team is under no illusions about the magnitude of the task ahead and has formed key international partnerships to move the project along. It has joined forces with a virtual cell consortium in the U.S. called EMC2, E-Cell in Japan and GlaxoSmithKlines' Biological Simulation division in the U.K. Together these groups formed the first international virtual cell alliance, sharing the work load and standardizing the way data will be collected.
At a conference in London this week, the partners will meet to share their progress to date. "We've just about completed the most comprehensive proteomic analysis of E. coli," said Ellison of CyberCell's contribution to a fact-sharing meeting to "kick start the development of the virtual cell."
Some of CyberCell's partners come from the most unlikely places - including Alberta's oil patch. Project CyberCell researchers have teamed up with Calgary-based Computer Modeling Group. The firm was formed three decades ago when it began to develop computer models of oil and gas fields for resource companies, simulating extraction scenarios that show geological changes as drilling progresses.
"They model space in 3-D and run scenarios over time, in other words four-dimensional modeling. We looked at their work and thought .gee, that's kind of what a virtual cell is going to look like.' " Ellison turns his attention to a computer screen image that resembles a schematic drawing of circuit boards in a VCR or some other household appliance. The circuit was designed by Gene Network Sciences, another one of CyberCell's partners, operating out of Ithaca, N.Y. GNS, among other things, has created a symbolic representation of what goes on in a cell.
"These guys construct networks of what gets turned into what. They've developed a whole language you can use to represent every kind of molecular event that goes on in a cell," he said of the drawing. "A lot of people say visualization of this is window dressing. It's not. We have to find ways to represent this so the human brain can understand it. We all prefer to look at a graph rather than stare blankly at rows and rows of numbers. The idea now is to get Computer Modeling Group together with the Gene Network Sciences people to build the circuit for E. coli in 4-D."
There are just too many things going on in a cell for researchers to get their heads around, says Ellison. "We need some kind of intellectual prosthetic to look for and recognize patterns. The virtual cell will make sense of extremely complicated molecular processes by computational modeling."
The job of Ellison and his team at the Institute of Biomolecular Design is to generate data about cells. "We will provide the biological relevance. We have an enormous advantage over other international efforts because each of our groups is so highly focused."
There are plenty of other efforts underway. In fact, they've helped spawn an entire industry. In 1998, investment analysts Frost & Sullivan predicted the market for bioinformatics equipment would be worth $2.2 billion by 2004. It has more recently estimated the market for proteomics instruments, worth an estimated $700 million in 1999, would top $5.6 billion by 2005.
The field promises not only to speed research findings but also to change the way research is managed. An inquiry as ambitious as Project CyberCell's means breaking down barriers between different disciplines.
"We have to be able to play nicely with computer scientists, mathematicians, physicists, biologists, biochemists and chemists. This in itself is a major challenge for us. This is not the way universities are structured. This project demands that fences be torn down."
Clearly, the creation of virtual life will not be achieved with a single discovery but will instead be the result of many smaller successes. Indeed, the dream itself was sparked by technological advancements.
Eillison's work in Project CyberCell is, in a way, the product of "a mid-life crisis."
"Science is so reductionist," said Ellison, a professor in the U of A's department of biochemistry. "You pick one or two aspects of cellular function and try to understand them in their entirety. Only in the last couple of years has the technology for being able to look at large numbers of proteins been around.
"So here we have this embarrassing amount of riches in data and the bottleneck is the interpretive engine. The limiting factor used to be our ability to collect data. The problem now is to interpret and gain insight from so much data."
Through the U of A's Institute for Biomolecular Design, Project CyberCell is already armed to do much of the work required. Its lab on the third floor of the Medical Sciences Building is vast. It houses an impressive collection of equipment: an X-ray area detector helps determine the structure of proteins in crystal form; a new, 600-mhz Nuclear Magnetic Resonance imager examines proteins in solutions, giving a more accurate physiological picture of their structure; three mass spectrometers, workhorses in functional proteomics, quickly identify proteins; and a robotic-equipped clean room helps researchers accomplish in hours what once required days.
Securing funding for the project has, unfortunately, been more difficult than Ellison and his colleagues hoped. It has already been awarded some $15 million for equipment from the Canada Foundation for Innovation, including $5.5 million announced in February. The Alberta government has announced $4.7 million to help secure the CFI funds, which are contingent on matching funds. Private partnerships are also being successfully pursued. IBM has just awarded Project CyberCell a $1 million Strategic University Research Grant. CyberCell has filed its first patent for a cell simulator and IBM is helping out so CyberCell's work is compatible with IBM's Blue Gene supercomputer.
But CyberCell is still about $1 million short of its current goals. "Some things will have to wait a little longer than we'd like them to," surmised Dr. Joel Weiner, Ellison's research partner.
Ultimately, though, Ellison thirsts for success, and longs for a day when his project and others like it have gone farther than most dare to dream, to the creation of a virtual human. "If we can do this with E. coli we can extend these methodologies to more sophisticated types of cells and create more complicated virtual cell types."