FIGURE 2 

Overview of one generation of the evolution algorithm. At each generation, the algorithm evolves the network collection (A). First, in each embryo, we solve for the expression pattern of a reporter protein E in the presence of a morphogen G as a function of position (B). The fitness, F, is defined as the number of jumps between high and low values of E. Half the networks of lowest fitness are discarded (C) and replaced by mutated copies of the fittest ones (D). This produces the starting network collection for the next generation (dashed arrow).