[gmx-users] Replica Exchange MD

[bharat gupta]

Dear GMX Users,

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://www.dropbox.com/s/dykxwwqulpfpw8x/REMD.png

I want to know whether, I can move forward with the large production runs
or not ??

Pls respond ....

-----
Bharat