Fig. 4

State and attractor spaces for two hypothetical boolean networks. State transition graphs of two rxncon networks (A, B) and their corresponding attractor spaces (C, D). In (A) and (B), simulation of the network from intermediate states (blue nodes) according to Boolean state transitions (black edges) causes the system to fall into a cyclic attractor (purple nodes). Addition of the ligand (green edges) or removal of the ligand (red edges) can cause the network to fall into a new attractor. In (A) and (C), addition of the ligand causes the system to fall into a single “with-ligand” attractor, and removal of the ligand causes the system to return to a single “no-ligand” attractor. This meta-attractor is visualized in the attractor space C as a self-consistent and reversible transition between two attractor nodes. In (B) and (D), removal of the ligand at different timepoints in the “with-ligand” attractor causes the system to fall into two distinct “no-ligand” attractors; this “branching” behavior is visualized in the attractor space (D) as a possible transition from a single attractor node to two different (a.k.a. Inconsistent with each other) attractor nodes both corresponding to the same “no-ligand” biological state of the system