Biological Informatics: The Virtual Cell: Modeling Cellular Processes

Biochemist John Carson uses The Virtual Cell to compare experimental and simulation results. Biologists are generating a vast amount of data on the molecular events that occur in the cell. Since a computer might the best tool for researchers to integrate all the information and sort out the complexities of a typical biological process, the next logical step would be to develop appropriate software for the job. A team at the University of Connecticut (UConn) Health Center in Farmington is attempting to do just that with a program called The Virtual Cell. The interdisciplinary effort involves biochemists, biologists, mathematicians, computer scientists, and physicists. Their goal is not to create a model in itself, or to create a system to model a whole cell, but to develop a tool for experimental biologists. Thus, the approach requires no user programming; instead, the user specifies biologically relevant abstractions such as reactions, cellular compartments, molecular species, and experimental geometry. In this way, a bench scientist can attempt to understand a cellular phenomenon by combining a collection of hypotheses and facts to create a model. Three main components are built into The Virtual Cell software. The program resides on a host computer accessible via the Internet. A biologically oriented user interface is capable of creating models, defining cellular geometry, specifying simulations, and analyzing simulation results. Another interface, mathematically oriented, allows one to work directly with mathematical problems and then perform simulations and analysis on those systems. By maintaining the physical laws and numerical methods in separate modular layers, The Virtual Cell is accessible to the experimental biologist and is a powerful tool for the theorist.1 The user inputs biochemical and electrophysiological data from experiments. That information is mapped to appropriate subcellular locations in two-dimensional or three-dimensional images obtained from a microscope. Behind the scenes, equations for chemical kinetics, membrane fluxes, and diffusion are generated and solved numerically. The results are again mapped to experimental images so that the cell biologist can fully utilize the familiar arsenal of image-processing tools to analyze the simulations. "One of the real strengths of this project is that we are keeping the development of the computational system closely connected to an environment where the experiments are being performed," emphasizes Leslie Loew, a physiologist at UConn Health Center. People in the laboratory gathering data, wanting to understand it, constantly interact with the people who are building the theoretical framework. Jim Schaff, a computer engineer on the project, predicts that The Virtual Cell will be available for general use fairly soon, perhaps in a few months. Schaff says the ultimate goal is to make the program easily accessible and intuitive. Researchers at UConn, and their collaborators at other universities, already are using it. Loew has successfully used the program to probe propagation of metabotropic calcium waves in neuroblastoma cells. The calcium wave, which can be quantitatively observed using a fluorescent indicator, preferentially initiates in the neurite in response to stimulation by bradykinin, a neuropeptide involved in the pain response. Loew focused on whether this phenomenon is a function of cellular geometry alone or it reflects spatial inhomogeneities in the distribution of specific receptors. The simulation of the calcium wave involved building a model from known information that elucidated the role of intermediates other than calcium, such as inositol 1,4,5-trisphosphate (IP3), diffusing through the cell. The simulation agreed with experiments regarding changes in the spatial distribution of free calcium, and predicted IP3 concentration and IP3 receptor open probability, which cannot be directly observed in vivo. Direct correlation of simulation results with experimental datasets was possible because the spatial organization of cellular structure was derived from a microscope image. The results suggested geometry alone is sufficient to explain wave initiation in the neurite. John Carson, a biochemist, is using The Virtual Cell to computationally simulate export of RNA from the nucleus to the cytosol and assembly of the exported granules. Though RNA trafficking is well-defined biochemically, there are many questions about the cellular pathway of the trafficking. Carson hypothesizes, "Maybe architecture itself is somehow steering this pathway and forcing it to only go in one direction. But there is no way you can test that experimentally." Carson is working with computer scientists to develop the computational models for microtubules and the nuclear pore. Once the various components are defined they can be placed in an area of cell geometry. Simulation results can then be compared to experimental data about how RNA moves out of the nuclear pore along microtubules. A prediction based on the model is, in a sense, most useful if it doesn't match the experimental details of the process, Loew points out. This indicates to the researcher that the elements of the model are inaccurate or incomplete. Although such negative results are not always publishable, they are a tremendous aid in refining understanding. If the prediction does match the experiment, it doesn't guarantee the truth of the model, but should suggest other experiments that can test the validity of critical elements. Ideally, it should also provide new predictions that can, in turn, be verified experimentally. The Virtual Cell team has overcome some of the skepticism biologists have toward computer modeling to win a $4 million National Institutes of Health grant. A significant sum of the grant is benchmarked for a computer system that can handle large information stores and many simultaneous users. As more people use the program, a valuable database of image information, chemical kinetics, and physiological information will be created so it can be used also as a teaching tool and reference. As to the dogma among experimentalists that cell biology is too complex to understand theoretically, Loew challenges biologists to a new way thinking: Only if biologists are open to using it will The Virtual Cell be successful in helping researchers deconvolute the complexity. Nadia S. Halim is a freelance science writer based in New London, Conn. J. Schaff, L.M. Loew, "The Virtual Cell," Pacific Symposium on Biocomputing, 4:228-39, 1999.

Biological Informatics: The Virtual Cell: Modeling Cellular Processes

Interested in reading more?

Become a Member of