Evolution of Evolvability in Gene Regulatory Networks
Figure 1
(A) Simulations are run on a 150×50 lattice for 6·105 time steps. The lattice harbors a population of genomes, where a genome is a linear chromosome of genes with binding sites. From a genome a Boolean threshold network is built. During each time step the network may update the expression level of the genes for 11 propagation steps. (B) The impact of several gene and binding site mutations is shown. The change in the genome and network topology is signaled by a red star. In a typical simulation the parameters are (per gene, binding site): gene duplication (dup) 2·10−4, deletion 3·10−4, threshold 5·10−6, binding site (bsite) duplication 2·10−5, innovation 1·10−5, deletion (del) 3·10−5, preference (pref) 2·10−5 and weight 2·10−5. See Model for an explanation on each type of mutation. (C) Typically the environment changes over time with a probability of λ = 3·10−4. The two evolutionary targets A and B determine which genes should be expressed (on) or inhibited (off). The result is four categories of genes; some should be always on, some should toggle their expression state and some should never be expressed. In a typical simulation, the target expression states are, from gene 0 to 19, A: 00011 11000 00000 11111 and B: 11010 01001 01100 01011.