A digital microbe confounds scientists' best predictions about its real-life counterpart's behavior. A virtual virus pinpoints new proteins that would elicit a strong immune response.

Those are just two examples of 'virtual cells', computer programs that mimic the complex biology of real organisms. Unlike pesky living cells, these cells multiply instantly, mimic any type of tissue and provide perfect reproducibility.

which we confirmed by experiment. Ronald Germain, US National Institutes of Health

Although cell simulations have been available for years, they were designed by theoreticians out of touch with what biologists need. They required users to learn obscure mathematical techniques, and few showed biological relevance. The new generation of simulators is far easier for biologists to use.

“This isn't the old days of theoretical biology where the theoreticians never really interacted with people in the lab,” says Les Loew, director of the Virtual Cell project at the University of Connecticut. “This is a tool for people in the lab to help them organize their experimental data and make predictions.”

The number of scientists using Loew's system has quickly grown to nearly 16,000. Virtual Cell is freely accessible online and sports a graphical interface that is strikingly similar to the blocky cartoons biologists draw on whiteboards. By drawing them in the Virtual Cell interface instead, scientists can produce sophisticated simulations of entire cellular systems.

“[The program] automatically translates the reaction diagram into a mathematical description, and then it automatically produces simulations from those,” says Loew. The simulations suggest experiments to test the model's underlying assumptions. Scientists can also share their models with others.

A similar program called Simmune helped create a virtual version of the common laboratory denizen Dictyostelium discoideum. The mold migrated toward virtual chemical attractants in a lifelike way, but displayed surprising molecular changes. The researchers hadn't looked for those changes before, but they confirmed that the model's unexpected predictions were correct.

“A lot of people who are not modelers think of modeling as curve-fitting, describing data that you already know, but that's not our goal,” says Ronald Germain, senior author on the work (PLoS Comput. Biol. 2, e82; 2006). “We predicted new biology which we confirmed by experiment.”

Using another program, scientists at the California-based La Jolla Institute for Allergy and Immunology created algorithms to predict the immune response against vaccinia virus, the basis of the smallpox vaccine. With the virtual immune system, the researchers predicted which structures, or epitopes, on the virus would bind most tightly to a cell surface protein that triggers antiviral immune responses. The predictions proved remarkably accurate, identifying several previously unknown viral epitopes (Nat. Biotechnol. 24, 817–819; 2006).

“I think this can narrow down the number of [vaccine] candidates that one wants to consider,” says senior investigator Alessandro Sette. Simulation could also help predict a vaccine's effectiveness without exposing people to the virus, an important concern for smallpox and other potential biological weapons.

The resurging interest in virtual cells is driven in part by an improvement in user friendliness, but biologists are also adopting more technical approaches. For example, Harvard University scientists have developed a new computer programming language for cell modeling.

As virtual cells evolve, testing their predictions will still mean confronting the challenges of live cells. “We've gone a long way in building models and making people excited,” says Jermey Gunawardena, a systems biology professor at Harvard, “but it's going to require some real breakthroughs at the level of single cell technology in order to really reap the benefits.”