Figure 11. Simultaneous estimation of hidden model states, maximal conductances and kinetic parameters in a two-compartment model of a vertebrate motoneuron (II).

Inference of maximal conductances, noise and kinetic parameters during smoothing. (A) Inferred parameters in the model using broad or narrow prior intervals and high or low levels of observation noise. Estimates were normalized according to Eq. 35. Parameter identification numbers are as in Table 2. The estimates in Ai were obtained using broad prior intervals (see Table 2). The maximal conductance (parameter #7) converged to zero and, for this reason, it is indicated with a red square. These estimates correspond to the results shown in Fig. 10. Estimates in Aii were obtained using narrow prior intervals for some of the parameters controlling the kinetics of ionic currents (see red dashed boxes) at either low () or high () levels of observation noise (see also Supplementary Figs. S4 and S5). (B) Inferred maximal conductances for sodium () and potassium () when narrow prior intervals and low levels of observation noise were used (circles in Aii). Notice the temporary convergence of the estimates (arrows) before jumping away towards their final values. (C) True (black lines) and inferred (red lines) activation and inactivation steady-states for the sodium and potassium currents (Ci) and the N-type and L-type calcium currents (Cii) and for the relaxation times for sodium inactivation and potassium activation (Ciii), when narrow prior intervals and low levels of observation noise were used (circles in Aii). In these simulations, , and the prior interval for was . The number of particles was in Ai and in Aii, B and C (see main text for further comments).