I have posted this query in a Gromacs forum, but got no reply. Can anybody give an advice on the following question?
I am trying to apply the REMD to study the difference between the folding free energy of a 24 residue peptide with two different conformations. As I am short of required number of processors for REMD in explicit water, I am trying my luck with implicit solvent. I have come across many papers that have used implicit solvent in their simulation studies. So, after various trails of temp. distributions, I selected some 15 replicas with the following temp. range:
Repl 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14
Repl T 250.0 264.0 279.0 295.0 312.0 329.0 348.0 367.0 388.0 410.0 433.0 458.0 484.0 511.0 540.0
Repl
Repl exchange interval: 1000
Here's the result for a trial run of 2ns per replica:
Replica exchange statistics
Repl 999 attempts, 500 odd, 499 even
Repl average probabilities:
Repl 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14
Repl .34 .34 .21 .27 .39 .41 .40 .37 .41 .42 .41 .43 .46 .44
Repl number of exchanges:
Repl 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14
Repl 166 155 116 130 184 201 193 185 212 216 200 208 223 225
Repl average number of exchanges:
Repl 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14
Repl .33 .31 .23 .26 .37 .40 .39 .37 .42 .43 .40 .42 .45 .45
The average exchange ratio for each replica lies between 0.3-0.4. Is it okay ? . As the value between 0.2-0.3 is considered to be good.
Here's the graph for replica_index , replica_temp and PE of replicas.
https://docs.google.com/file/d/0B6ehLXK0eP7sQWRNb010MXNLM28/edit
I want to know whether, I can move forward with the large production runs or not ??