Hey folks,

My acceptance ratios during a 32-replica 310-600K REX run (14-mer peptide in a 40 A^3 TIP3P box with 5 Na+ and 11 Cl- ions) are too low for lower temperatures and too high for higher temperatures:

0 1 90 1144 7.87
1 2 106 1314 8.07
2 3 103 1014 10.16
3 4 105 1087 9.66
4 5 112 1118 10.02
5 6 109 1079 10.10
6 7 102 1064 9.59
7 8 122 1106 11.03
8 9 128 1061 12.06
9 10 120 1082 11.09
10 11 149 1092 13.64
11 12 133 1077 12.35
12 13 136 1076 12.64
13 14 146 1099 13.28
14 15 163 1080 15.09
15 16 157 1101 14.26
16 17 161 1086 14.83
17 18 191 1087 17.57
18 19 197 1100 17.91
19 20 219 1095 20.00
20 21 184 1072 17.16
21 22 183 1092 16.76
22 23 194 1098 17.67
23 24 209 1063 19.66
24 25 231 1103 20.94
25 26 225 1094 20.57
26 27 241 1070 22.52
27 28 250 1097 22.79
28 29 219 1044 20.98
29 30 303 1286 23.56
30 31 278 1174 23.68

My aarex.pl input looks like this:

aarex.pl -hosts nodefile.txt -n 5010 -temp 32:310:600 \
-par archive,initruns=0,equilruns=10,natpdb=lka14_sys_equil_310.pdb \
-mdpar param=27,shake=1,shakemode=all,dynsteps=500,dyntstep=0.002 \
-mdpar boxshape=cubic,boxsize=40.0 \
-mdpar cuton=10,cutoff=12,cutnb=14 \
-ens rex -ensdir data -elog energy.log -charmmlog charmm.log \
lka14_sys_equil_600.pdb

... and my temperatures are distributed by ReXServer.pm as follows:

cond 310.000000
cond 316.674400
cond 323.492500
cond 330.457400
cond 337.572300
cond 344.840300
cond 352.264900
cond 359.849300
cond 367.597000
cond 375.511400
cond 383.596300
cond 391.855300
cond 400.292100
cond 408.910500
cond 417.714500
cond 426.708000
cond 435.895200
cond 445.280200
cond 454.867200
cond 464.660700
cond 474.665000
cond 484.884700
cond 495.324500
cond 505.989000
cond 516.883100
cond 528.011800
cond 539.380000
cond 550.993100
cond 562.856200
cond 574.974700
cond 587.354100
cond 600.000000

Does anyone know of a "fix" for this temperature/energy distribution issue (or am I doing something wrong)?

Thanks!

Change the temperature spacing-- make the bath T values closer together at lower T, and further apart at higher T.

How do you override the exponential distribution set in aarex? Is it as simple as creating a new list and putting it in rexserver.pl? The '-temp' switch is required, along with the high and low temperatures, so we are not at all sure how to do this.

Cheers

Mike

To assign specific temperatures, you can use the "-condfile" option and provide a condition file containing the temperatures. Here is a simple example (8 replicas using the same starting structure):


aarex.pl -n 10 -mdpar param=27 -charmmlog logfile -log logfile -condfile conditions structure.pdb

Where the condition file "conditions" contains a list of the temperatures:

300
301
302
303
304
320
330
340

Thus, each of the 8 replicas is assigned to one of the listed temperatures (you can verify this in your CHARMM log file by grepping "Reference temperature"). You will notice that the "-temp" option is omitted in this case and the number of replicas is implicit in the number of temperatures listed in the condition file.

Hope this helps.