PT  - JOURNAL ARTICLE
AU  - M. L. Blinov
AU  - J. C. Schaff
AU  - D. Vasilescu
AU  - I. I. Moraru
AU  - J. E. Bloom
AU  - L. M. Loew
TI  - Compartmental and spatial rule-based modeling with Virtual cell (VCell)
AID  - 10.1101/146225
DP  - 2017 Jan 01
TA  - bioRxiv
PG  - 146225
4099  - http://biorxiv.org/content/early/2017/06/05/146225.short
4100  - http://biorxiv.org/content/early/2017/06/05/146225.full
AB  - In rule-based modeling, molecular interactions are systematically specified in the form of reaction rules that serve as generators of reactions. This provides a way to account for all the potential molecular complexes and interactions among multivalent or multistate molecules. Recently, we introduced rule- based modeling into the Virtual Cell (VCell) modeling framework, permitting graphical specification of rules and merger of networks generated automatically (using the BioNetGen modeling engine) with hand-specified reaction networks. VCell provides a number of ordinary differential equation (ODE) and stochastic numerical solvers for single-compartment simulations of the kinetic systems derived from these networks, and agent-based network-free simulation of the rules. In this work, compartmental and spatial modeling of rule-based models has been implemented within VCell. To enable rule-based deterministic and stochastic spatial simulations and network-free agent-based compartmental simulations, the BioNetGen and NFSim engines were each modified to support compartments. In the new rule-based formalism, every reactant and product pattern and every reaction rule are assigned locations. We also introduce the novel rule-based concept of molecular anchors. This assures that any species that has a molecule anchored to a predefined compartment will remain in this compartment. Importantly, in addition to formulation of compartmental models, this now permits VCell users to seamlessly connect reaction networks derived from rules to explicit geometries to automatically generate a system of reaction-diffusion equations. These may then be simulated using either the VCell partial differential equations (PDE) deterministic solvers or the Smoldyn stochastic simulator.