Hi All, I am running a set of umbrella sampling simulations to study the movement of a small ligand molecule through a beta barrel nanopore kind of protein. In order to calculate the free energy of the ligand moving from one end of the protein to the other, I am running several umbrella sampling simulations with different harmonic restraint potential force constants (high for windows where reaction coordinate values are deviating easily). I would like to perform WHAM analysis to get the corresponding free energy of the process as a function of the reaction coordinate. I am getting a pmf plot with very high free energy barriers of ~1200 kJ mol-1. It does look like weird sigmoid curve. The reaction coordinate is the z-component of the ligand -protein com-com distance vector and it ranges from -3.0 nm to +3.0 nm. I have tried gmx WHAM with the -cycl flag. This assumes the reaction coordinate to be periodic, having same pmf value at -3 and +3 nm. Using -cycl flag with gmx wham gives me a decent looking profile where both the ends (pmf at -3 and +3 nm) are at zero. However, when I run a bootstrap error calculation with nBootstrap = 200, I see the error bars associated with the pmf is pretty large on the order of 40-50 kJ mol-1 in some cases. I have run 10 ns simulations for total 121 umbrellas. So, I have two questions, for the purpose of studying the transfer of the ligand through the protein as function of the z-distance, is using gmx wham with periodic pmf a better alternative than the non-periodic pmf; what does such large error bars associated with pmf actually mean, do I need to increase my sampling even further to lower the error bars? Any help will be much appreciated, thank you