Violin plots describing the estimated probability distributions of parameters. (A–E) q_i (A), V_i (B), W_i (C), x_i (D), and ln(ν_i) (E). The best fit solution (X) exhibits the smallest error out of 1.5 ⋅ 10⁶ samples distributed in three replicas and corresponds to the model predictions presented in Fig. 7 (black lines). The mean and median of each parameter are shown as red and black horizontal lines, respectively. The vertical blue line indicates the 95% CI for each parameter. The shaded areas depict the posterior probability density function for each parameter obtained by the Markov chain Monte Carlo sampling smoothed by a kernel density estimator (Hoffmann, 2021). The computational model yielded a highly constrained set of parameters describing simultaneously datasets from both WT and R174W mutants (Table S1). Note that, since VSD-III and VSD-IV activation kinetics are extremely fast, the actual relative position in the electric field of the transitions energy barrier (x₃ and x₄) has little influence on the goodness of fit, as illustrated by the vertical spreading of these parameter distributions (D).