Stochastic simulation of particle dynamics
Numerous frameworks can be employed to create computational models of biological systems. Which framework to chose is heavily dependent on the particular problem one is interested in solving. Often there are trade-offs between what is realistic in terms of the underlying biology of the model and what is realistic in terms of the resources available to the researcher, such as computational power and biological data.
We are currently developing a simulation software in our laboratory to create and run simulations for chemical reaction-diffusion systems at the mesoscopic scale. Any biological entity of interest, such as a protein molecule, is represented by a software object which propagates in continuous simulation space according to Brownian Dynamics algorithms. Several diffusion spaces can be defined (static, free diffusion, within membrane, above membrane, blow membrane). Brownian motion describes movement of the particles in two or three dimensions. The mobility of freely diffusing particles is calculated according to Stokes' law, whereas the mobility of particles in the membrane is calculated according to a model by Saffman and Delbruck.
Simulations occur in a three dimensional simulation volume, the boundaries of which can be reflective, toroidal or absorbing. The simulated entities contain crude geometric information such as the concept of protein domains and active sites and the relative position of reactive surfaces to the entities centre of mass. Entities can bond, forming transient aggregations (clusters), while still maintining their identities. The simulation allows for zero-order (creation), unimolecular (e.g. conversion, destruction) and bimolecular (e.g. association) reactions to occur in solution, according to stochastic algorithms based on the set of Smoldyn algorithms. This molecular resolution allows us to address questions found in the field of molecular interactions and signal transduction of biological systems, where local concentration effects and the geometry of interactions are often important
The current version and example model files can be downloaded here.
The source code is available here.
The Simulation software is distributed under the General Public Licence.
The original idea for the simulation software was based on the Abstracted Protein Simulator, developed at Edinburgh University. While much of the original code dealing with object rendering still persists in the simulator, the overall simulation engine and the code dealing with simulation volume has been entirely rewritten
This simulator is developed by Dominic Tolle at the time PhD student in the Le Novere Lab then located at the EMBL-EBI. The author is grateful to Dan Mossop, Fred Howell (fwh at inf.ed.ac.uk) and Eilidh Grant (egrant1 at inf.ed.ac.uk) for their initial work.