Figure 3.

GRN coarse-graining. (a) The GRN as a Boolean network. Note Isl1 has further projections into the larger cardiac regulatory network, but since these are not represented, it is instead given a self-loop as is traditional in Boolean analysis for nodes without outputs. (b) The expanded state-space wherein each node is a binary string of exogen_canWnt_II, Foxc1_2, Fgf8, CanWnt, Isl1. (c) Using a greedy algorithm that groups different sets of nodes together the possible partitions can be explored in a search for only those groupings that improve the EI. (d) Once the appropriate groupings are identified the network can be represented in its dimensionally-reduced format. Here the macro-nodes μ₁ and μ₂ are constructed in the simplest way possible, as a coarse-grain.