Computational biology reached a milestone last week. Scientists at Stanford University and the J. Craig Venter Institute presented the first software simulation of an entire living organism, albeit an extremely simple one: the single-cell bacterium Mycoplasma genitalium. The simulation models M. genitalium’s metabolic functions and all of its 525 genes (for comparison, multicellular organisms can have tens of thousands of genes).
Stanford’s project website opens with a time-lapse video demonstrating how the model’s various parameters change throughout a simulation run. The website also offers a comprehensive M. genitalium knowledge base and other resources.
Another project archive is hosted at SimTk, a software repository for physical simulations of biological systems. Here you can find the entire source code and training data for the simulation (requires Matlab to run).
Also on SimTk, the unassumingly titled Supplementary Text (PDF) describes how the simulation works internally, complete with pseudocode algorithms and data structures.
The model was based on over 900 research publications for M. genitalium and is supposedly precise enough to predict new behavior. While this project is a remarkable achievement, extending it to more complex organisms is going to be quite a challenge: a 128-core Linux cluster needs 10 hours and produces 0.5 GB of data for one simulated cell division!
Those interested in biological simulations should also check out another project that operates on a cellular rather than a molecular level. OpenWorm aims to reproduce the popular nematode Caenorhabditis elegans. The simulation is a work in progress but you can already admire a multi-layered 3D model in the OpenWorm Browser (requires WebGL).