Fig 8. Example of graph transformation with BNG reaction rules.

This example shows details of how BNG reactions are implemented in the libBNG library [19]. Reactants are defined with molecule types A(c0∼R∼S,c0∼R∼S) and B(c1∼U∼V,c2∼X∼Y) where A and B are names of the molecule types, c0 is a component of A that can be in one of the states R and S, and similarly c2 and c3 are components of B. (A) is the example reaction rule, (B) are example species reactants and products in the BNGL syntax, and (C) shows a graph representation of the rule in (B). Application of the rule is done in the following steps: 1) a mapping from each molecule and each component from reactant patterns (E) onto reactants (D) is computed (dotted arrows). If the state of a component is set in the pattern, the corresponding reactant’s component state must match. The next step 2) is to compute a mapping of the rule product pattern (F) onto reactant patterns (E). The difference between the reaction rule product pattern and the reactant patterns tells what changes need to be made to generate the product. In this example, a bond between A’s component c0 with state R and B’s component c1 is created. The state of A’s component c0 is changed to S. Once the mappings are computed, we follow the arrows leading from the reaction rule product pattern (F) to reactant patterns (E) and then to reactants (D) and 3) perform changes on the reactants resulting in the product graph (G). Each graph component of the product graph is a separate product and there is exactly one product in this example.