Figure 3.
For the two possible pathways from wild-type protease to the 10/90 double mutant, we computed the conditional Ka/Ks values for each mutation conditioned on the presence or absence of the other mutation (shown as numbers next to each edge in the figure). For example, in the absence of the 10 mutation, the 90 mutation shows strong positive selection in both the Specialty and Stanford-Treated datasets, but was negatively selected in the Stanford-Untreated dataset. Since the steady-state speed of a multistep path is determined by its slowest step, we highlighted the rate-limiting step in each path (boldface). For example, in the Specialty dataset, the steady-state rate of the upper pathway appears to be ∼10-fold faster than that of the lower pathway. (a) Specialty dataset, (b) Stanford-Treated dataset and (c) Stanford-Untreated dataset.