The idea behind non-equilibrium FEP is to extract snapshots from equilibrated end states for both lambda 0 and lambda 1. You then use these snapshots as starting configurations to run short transition simulations, pushing lambda from its one state to the other.
It is my understanding that accurate estimations of ddG require uncorrelated samples, and therefore can we make the inference that supplying multiple correlated samples is redundant, when a single one of these samples could be used instead?
If I were to extract only the uncorrelated snapshots from my EQ end state simulation, and only run transition simulations with these, would I be reducing my computational overheads? I.e., I could reduce the number of transitions I'm running by only running uncorrelated ones. If so, is it possible to extract uncorrelated frames using pymbars autocorrelation functions?
The idea behind non-equilibrium FEP is to extract snapshots from equilibrated end states for both lambda 0 and lambda 1. You then use these snapshots as starting configurations to run short transition simulations, pushing lambda from its one state to the other.
It is my understanding that accurate estimations of ddG require uncorrelated samples, and therefore can we make the inference that supplying multiple correlated samples is redundant, when a single one of these samples could be used instead?
If I were to extract only the uncorrelated snapshots from my EQ end state simulation, and only run transition simulations with these, would I be reducing my computational overheads? I.e., I could reduce the number of transitions I'm running by only running uncorrelated ones. If so, is it possible to extract uncorrelated frames using pymbars autocorrelation functions?