Preferentially coupled metabolite pairs in E. coli. The metabolite pairs in E. coli that are greater than 99% likely to be preferentially coupled in a non-random fashion are summed by row/column and binned. The ordering of the rows in this figure is identical to figure 3, proceeding from most to least connected metabolites from left to right. Each preferential coupling is counted twice, once for each metabolite involved. The most connected metabolites occupy the bins on the left, and clearly account for most of the preferential coupling detected.