I am studying an enzyme where the closing of a loop should happen upon ligand binding to the active site. In unbiased simulations, even after 300 ns of simulation time the loop does not close the reaction center. Thus, we realized that the open-closed conformational change may have a high free energy barrier, inaccessible by the unbiased trajectory. Thus, we performed well-tempered metadynamics (WTM) simulations to study that event.
In WTM simulations, we see the opening and closing of the loop almost five to six times within 300 ns time span. The free energy associated with this event has barriers of only few kJ/mol (almost flat surface). Now, given these low values of the free energy barriers, I would expect to observe the open-closed transitions in the unbiased simulations. However, we do not see that transition in the unbiased trajectories.
So, my questions are,
1) Are the collective variables used here inappropriate? (Although I see the open-closed
transition a many times in the metadynamics trajectory.)
2) Is the free energy not converged? (plotting hills heights a function of time, decays close to zero after 200-250 ns)
3) How would I make sure that the free energy is converged? (block averages using -stride in plumed) How would I decide the blocks from the COLVAR plot?