Figure 5

a) Mean Hamming distance between viable genotypes. The horizontal axis shows the number n of reactions and the vertical axis shows the mean Hamming distance between random genotypes in the set V(n), and for a comparison, in Ω(n). Data are shown for three different environments (glucose, acetate and succinate). For each V(n), we computed the data using all pairwise distances between our 1000 sampled genotypes. The dashed black curve shows mean Hamming distance for the case of Ω(n), which can be computed analytically and is given by 2 n(N-n)/N. The viability constraint affects the mean Hamming distance significantly only at low values of n. We also show the maximum Hamming distance in Ω(n) (dotted black line). b) Maximum Hamming distance between viable genotypes. The horizontal axis shows the number n of reactions and the vertical axis shows the maximum Hamming distance between genotypes in the set V(n) and, for comparison, in Ω(n). Data are shown for three different environments. For each V(n), we computed the data through a modified two-headed MCMC approach (see text for details). The dotted black curve shows the maximum distance in Ω(n) which can be computed analytically and is given by 2n for n<N/2 and 2(N-n) for n>N/2. Again, the viability constraint affects the maximum Hamming distance significantly only at low values of n.