Membrane Permeability from Conventional MD Simulations: Counting Transitions vs. Bayesian Analysis
The permeation of small molecules through membranes can presently be observed in conventional (i.e., non-enhanced) molecular dynamics simulations. This contribution focuses on three important aspects of such calculations. (1) The advantages and disadvantages of calculating permeability by direct counting of transition events versus Bayesian analysis based on the inhomogeneous solubility diffusion model. (2) A new Python/C++ tool that speeds up a previous implementation of the Bayesian analysis by two orders of magnitude and allows permeabilities to be extracted in a matter of seconds from a previously generated trajectory. (3) Simulated and permeabilities of water, oxygen, and ethanol through various homogeneous bilayers. The results fall short of the experimental values, clearly demonstrating the requirement for accurate polarizable force fields.