The Biomolecular Network Simulator (BNS) uses the Gillespie stochastic algorithm or a tau-leaping algorithm to simulate a system of biochemical reactions. The BNS code consists of a combination of MATLAB and C-coded modules that allow the use of interactive features and visualization tools, while achieving high speed for the computationally intensive part of the calculations. Since the biochemical reactions are stochastic in nature, the user must run thousands of simulations to characterize the "ensemble" behavior of biological systems. Therefore, the BNS code is parallelized with the MPI library, allowing users to run simulations on multiprocessor computers. The BNS software can be run on any computer platform where MATLAB 6.5 or newer is installed.

BNS consists of two sets of tools: one for the simulation of chemical reactions and another for the analysis of the results. The graphical user interface of BNS allows users to easily input model and simulation parameters and to select the analysis method. Multiple types of post-simulation analyses are available.

Model Input and Simulation Setup

• The input reaction model can be specified in BNS or SBML format
• Graphical user interface for: model selection, parameter set selection, modification and saving, setting simulation parameters, specifying state variables to monitor and to plot, selecting number of simulation runs to perform, selecting whether to run simulations on PC or HPC hardware, and automatically generating batch script for HPC

Simulation Engine

• Executes stochastic simulations using exact Gillespie and approximate tau-leaping algorithms
• Supports parallel CPU operations for performing multiple runs simultaneously
• Saves the results at user-defined time intervals (snapshot data) and provides an event log of every reaction event that occurs

Data Monitoring

• Plots state variables (numbers of molecules) vs time during simulation
• Plots multiple state variables on a single graph for comparison and interpretation

Analyses

• Provides a user friendly graphical interface for setting up analyses workflows
• Plots the exact behavior of selected state variables (number of molecules) vs time
• Computes time-weighted averages of each compound over a specified time interval
• Computes and plots histogram of reaction rates and time-averaged reaction rates
• Conducts sensitivity analysis
• For a model system containing 250 compounds and 280 chemical reactions, BNS can typically perform 1,000 simulations per CPU/hour in an HPC environment.

The software was developed in collaboration with Dr. J. Frazier (Air Force Research Lab) and Dr. B. Foy (Wright State University). It is currently deployed on the falcon machine at the Air Force Research Lab Major Shared Resource Center.