STEPS (STochastic Engine for Pathway Simulation) wasdeveloped as a tool for simulating reaction-diffusion sys-tems, such as the signaling pathways involved in synapticplasticity. We aim to provide the tools to allow investiga-tion of the effects of morphology and stochastic fluctua-tions in such systems, since it is widely believed suchfeatures may play an important role in biological function[1]. STEPS is written almost entirely in C++ for high per-formance, but maintains a user interface in Python toallow for greater interactivity and control. The softwareruns on various platforms, including Unix, Mac OS X andWindows. By writing relatively straightforward Pythonscripts the modeler can create the molecular species in thesystem, set up the reaction-diffusion kinetics, define thegeometry of the system, set the initial conditions, importa 3D tetrahedral mesh, run and control the simulationand visualize and/or save the simulation results.STEPS currently includes three solvers, allowing the userto choose the most efficient solution that can accuratelydescribe their model:-WmDirect: Stochastic solver, based on the GillespieDirect Method [2].Recommended for systems that can be assumed to bewell-mixed and where any species is in a sufficiently lowconcentration that the probabilistic nature of the systembecomes significant and a stochastic description is neces-sary [1,3]. The modeler can adapt their Python script torun the simulation a number of times to see the stochasticfluctuations in the system.-WmRK4: Deterministic solver, based on the fourth-orderRunge-Kutta method.Recommended for systems that can be assumed to bewell-mixed and where molecular populations are largeenough that system behavior can be approximated bysolving classical chemical kinetics equations. This is usu-ally the quickest solver as the simulation need only be runonce.-TetExact: Stochastic solver extended for simulating diffu-sion by implementing a 3D tetrahedral mesh.Allows for the investigation of the effects of morphologi-cal aspects such as concentration gradients and chemicalcompartmentalization. This solver requires constructionof a high-quality computational tetrahedral mesh that canadequately represent the relevant morphological features.STEPS currently supports meshes generated by TetGenhttp://tetgen.berlios.de, which is freely available forresearch use. In addition, we aim to implement mesh-gen-eration in STEPS in the future based on the algorithm in[4]. These meshes can then be used for simulation byextending Gillespie's Direct Method [2] to deal with spa-tial processes. Each voxel is treated as a well-mixed vol-ume in which reactions can occur and are coupled to theirneighboring voxels by diffusive fluxes and to neighboring
Read full abstract