[gmx-users] biasing potential replica exchang for polymer

[leila salimi]

Dear gromacs users,

I am using  16 replicas with different dihedral potential for peptide in
explictit solvet. for 20 ns I got this result:

Replica exchange statistics
Repl  9999 attempts, 5000 odd, 4999 even
Repl  average probabilities:
Repl     0    1    2    3    4    5    6    7    8    9   10   11   12   13
  14   15
Repl      .35  .30  .35  .38  .44  .25  .58  .63  .22  .32  .38  .33  .32
 .37  .33
Repl  number of exchanges:
Repl     0    1    2    3    4    5    6    7    8    9   10   11   12   13
  14   15
Repl     1751 1489 1761 1931 2166 1281 2883 3134 1105 1582 1856 1629 1604
1869 1676
Repl  average number of exchanges:
Repl     0    1    2    3    4    5    6    7    8    9   10   11   12   13
  14   15
Repl      .35  .30  .35  .39  .43  .26  .58  .63  .22  .32  .37  .33  .32
 .37  .34

I think the exchange probabilities between replicas is OK. Am I right?

I have another question, I want to simulate my system for 40 ns
for equilibration and then 400 ns for data analysis. After 40 ns, Should I
use the last configuration of each replica and make the new tpr file for
each replica or I can use one of the last configuration of replicas after
40 ns and make the new tpr file for each replica (with the same gro file)?

I think when I want to equilibrate each replicas, then I have to use the
last configuration of each replicas and continue the simulation.

I am not sure that if I start replica exchange MD with different
trajectories of the system, then I will get the correct exchanges between
replicas.

Sorry that my questions might be very basic.

Regards,
Leila