Virtual Cells Could Speed Drug DiscoveryBy M.L. Baker | Posted 06-06-2005
Researchers could get results in less time using virtual cells rather than real ones. They also could avoid working with dangerous pathogens.
And so far, digital experiments predicting how bacteria will swim in response to environmental cues have produced results similar to those from experiments using real bacteria, for both single cells and bacteria populations, according to a paper published this month in the journal Bioinformatics by Philippe Cluzel, Thierry Emonet and others at the University of Chicago's Institute for Biophysical Dynamics.
The "digital cells" are virtual versions of the species E. coli, the microbial equivalent of a laboratory rat. These digital bacteria can sense likely food molecules in a virtual environment. They can transmit biochemical signals to their flagella, whiplike structures that bacteria use for locomotion, correctly mimicking how the bacteria will move.
Using the digital cells, Emonet found that levels of a particular protein could make a bacterium more or less likely to respond to environmental cues. Follow-up experiments on real bacteria showed the same thing.
The advantage of this computer simulation is that it can simultaneously model the activity of the molecular machinery inside a cell, the bacterial cell itself and a group of several bacteria. Though other simulations can model these individually, the Chicago researchers say they know of no other software that can model all three scales at once.
"With AgentCell, we can simulate the behavior of entire populations of cells as they sense their environment, respond to stimuli and move in a three-dimensional world," Emonet said.
The model may help predict when bacteria cells will act individually or as a group. That could be important for understanding biofilms. These mats of bacteria can cause infectious kidney stones, endocarditis (infection of heart valves), and cystic fibrosis lung infections. Biofilms also may be useful for tasks such as sewage treatment.
In silico techniques abound for showing how potential drug molecules may bind to proteins on or within a cell. But these do not show how that binding could affect what a cell does.
For example, while a computer program might predict that a potential cancer drug binds to a protein believed to encourage cell growth, the program cannot predict whether the drug disables the protein or whether disabling the protein could keep the cell from growing.
The goal of the software is predicting how a cell's environment and interactions with other cells affect internal biochemical fluctuations, and how those fluctuations affect a cell's behavior.
Right now, the bacteria have only a sensory and reaction system, but more modules, such as cell division, could be introduced. The developers hope other researchers will improve the model, modifying code and adding new capabilities.
Information for obtaining the code is available here.
Check out eWEEK.com's for the latest news, views and analysis of technology's impact on health care.