Research
Research Issues. How do you simulate an entire creature on the metabolic scale? Being able to predict how certain classes of mice react will be a novel and groundbreaking new method for drug discovery. The calculations made in virtual cells and virtual mice can be transferred to synthetic biologists who are working to build a cell and eventually a mouse "from the ground up."
While differences between humans and animals ultimately dictate the usefulness of experimental animal models, much has been learned from experiments in animals where the variables can be rigorously controlled. The laboratory mouse, representing hundreds of distinct strains, is a rich genetic resource made increasingly more valuable through emerging technologies that manipulate the mouse genome. An advantage of the known genetics of distinct mouse strains is that information at the gene can be followed through to a phenotype. An approach used in recent years to find genes related to a phenotype is quantitative trait loci analysis, which relies on measurable phenotypic differences between inbred mouse strains (Moore and Nagle, 2000). This strategy uses the segregation of one or more traits of interest that are followed by crossing inbred strains that differ significantly in the measured phenotype. The backcrossed or intercrossed progeny are then phenotyped for the trait and genotyped for a series of genetic markers spanning the genome that are polymorphic between the parental strains. While valuable, this approach is time consuming, dependent on extensive genetic manipulation, and requires the ability to assess quantitatively a phenotypic trait. In silico simulation of the mouse will enable researchers to understand complex phenotypes without the need for extensive genetic breeding and manipulation.
An additional advantage of studies in mice is that many diseases can be recreated in mice using gene knockout technology in embryonic stem cells. In recent years, mouse models of disease have become commonplace because the information gained from in vivo study of biochemical pathways is invaluable. Finally, unlike living humans, every part of the living laboratory mouse can be accessed relatively easily by sensors or laboratory analytical procedures. This ability is vital for testing the predictions of the virtual mouse simulation and providing "quality control."
Related Research. The Virtual Cell project at the University of Connecticut Medical Center is producing a database of known reactions at the cellular level. We can use the database instead of computing similar reactions. In the long term, we would like all calculations at the grid nodes to be database lookups.
A universal computational framework for modeling biological processes in a cell has been developed with NIH funding. This framework is called the “Virtual Cell”. The Virtual Cell Portal is available online at http://www.nrcam.uchc.edu. We intend to build upon this software framework by establishing the Virtual Mouse first as a collection of virtual cells in energy and mass communication with one another, and then by establishing emergent properties of this larger system. The Virtual Mouse will be more than a collection of Virtual Cells, but it must start with a collection of Virtual Cells. Wherever possible we intend to provide entry of mathematical equations that describe a model using the established declarative language (Virtual Cell Mathematics Description Language, VCMDL).
The Virtual Cell is intended as an instrument for experimentalists. In the Virtual Cell, models are assembled from electrophysiological and biochemical data mapped to suitable subcellular positions in images acquired from a microscope, or other analytical technique with appropriate spatial resolution. Membrane fluxes, chemical kinetics, and diffusion are therefore joined and the ensuing equations are solved numerically. Subsequently, the results are mapped back to experimental images to enable the scientist to exploit fully the latest battery of image processing tools to analyze the simulations. Using the “Virtual Cell” necessitates an operational explanation of the term “model”. In this context, the concept is best understood as a rewording of the scientific method. A model is just a compilation of hypotheses and facts that are assembled in an effort to comprehend an event. The selection of which hypotheses and facts to gather and the way in which they are pulled together, themselves comprise further hypotheses. A prediction founded on the model is often most useful when it does not correspond to the experimental facts of the process. This is operationally “watching” the virtual mouse “die“, when its physical counterpart continues to thrive. In such cases, it then clearly informs us that the essentials of the model are erroneous or incomplete. Even though such negative results are not always publishable, they are a wonderful help in refining our comprehension. When the prediction does agree with the experiment, it cannot certify the veracity of the model, but it ought to recommend other experiments that can check the validity of crucial elements. If possible, it should also present new forecasts that can then be demonstrated experimentally. The Virtual Cell itself is not a model. The software and framework are simply provided as a tool for scientists to generate models and to make predictions from models through simulations. The underlying mathematics, physics, and numerics included to date have been thoroughly validated. The results of simulations from the Virtual Cell are offered in a format that can be analyzed using procedures similar to those
Research Approach. A very large grid will be laid over the physical domain of a mouse. At each grid node is a chemical reaction that can be simulated using a program like Gaussian or NWChem unless the reaction is already known and in a database. Between nodes is a simple heat and mass transfer problem that is solved between time steps. The biomedical and library scientists will provide empirical and hand-curated data to the computational modelers, who will run simulations and produce results, and the biomedical scientists will validate the results for the modelers, which will repeat. This is the “simulation cycle.” A multiscale (or multigrid) approach will significantly speed up the calculations.
One of the most interesting and important things about the virtual mouse initially will be watching it "die" (in the early stages of the project, this class l/class 2 cellular automaton type behavior might happen much more frequently than it does with a physical mouse). Determining why the virtual mouse dies will be very informative and a key component in improving the virtual mouse model.
The long term goal of research aimed at creation of a Virtual Mouse is to use computer simulations to understand the basis of disease. Developing computer simulations for blood pressure in normal mice will be complemented in our approach by extending the simulations to mice with various disorders of the Metabolic Syndrome. The Metabolic Syndrome, highly prevalent in the US and in the Commonwealth of Kentucky, encompasses the diseases of obesity, diabetes, dyslipidemia, and hypertension. Because the Syndrome clusters together several diverse diseases, it epitomizes the need for interdisciplinary approaches to deal with the complexity of the disorder. Successful development of computational models that predict the complex behavior of blood pressure, a physiologic response heavily influenced in the Metabolic Syndrome, will lay the groundwork for the creation of computer simulations describing complex physiology in the presence of disease.