Dear Amber community
I'm trying to run a restrained simulated annealing of a fluornie modified RNA double helix using a generalized Born implicit solvation model, however the calculations often fail. After examining each step of the calculation I can see that the molecule is split through a plane into two parts (bonds don't actually break) which start drifting away from each other extending the bonds until in the end a long linear arrangement of atoms is left. The molecule splits every time through a different plane. I tried running the calculation on different machines with different architectures on parallel and serial versions of sander. The starting structure is minimized and prolonged equilibration makes no difference. I tried with the parmbsc0 and parm99 force fields with same results.
I have tried combinations of all kinds of parameters (dt, ntt, ntc, ntf, igb...) and am slowly running out of ideas. I found that the higher the temperature the more likely it is the calculation will fail. If temperature doesn't exceed 300K the calculation usually ends well, if the temperature reaches 1000K more than 90% of the calculations fail. However, I didn't run into any problems for same calculations in vacuum (at any temperature). Running the calculation with igb>0 on a different unmodified RNA molecule ends without problems even at 1000K. Looking at this the problem appears to be in the modified residues, but I have no idea where to start looking. Any suggestion would be greatly appreciated.
Here is one of my input files:
&cntrl
nmropt=2,
ntpr=200, ntwx=200,
nstlim=60000, dt=0.0005,
tempi=300.0, temp0=300.0, ig=RANDOM_NUMBER, ntt=1,
ntc=2, tol=0.0005, offset=0.13,
pencut=-0.001,
ntf=2, ntb=0, cut=20.0, igb=1,
/
&wt type='TEMP0', istep1=0,istep2=5000,value1=300.0,
value2=1000.0, /
&wt type='TEMP0', istep1=5001,istep2=35000, value1=1000.0,
value2=1000.0, /
&wt type='TEMP0', istep1=35001,istep2=46000, value1=1000.0,
value2=100.0, /
&wt type='TEMP0', istep1=46001,istep2=60000, value1=100.0,
value2=0.0, /
&wt type='TAUTP', istep1=0,istep2=5000,value1=0.4,
value2=0.4, /
&wt type='TAUTP', istep1=5001,istep2=35000,value1=1.0,
value2=1.0, /
&wt type='TAUTP', istep1=35001,istep2=36000,value1=1.0,
value2=1.0, /
&wt type='TAUTP', istep1=36001,istep2=60000,value1=0.2,
value2=0.2, /
&wt type='REST', istep1=0,istep2=15000,value1=0.1,
value2=1.0, /
&wt type='REST', istep1=15001,istep2=60000,value1=1.0,
value2=1.0, /
&wt type='END' /
LISTOUT=POUT
DISANG=rst_allV2
And the output of the steps where the problem occurs:
NMR restraints: Bond = 66.558 Angle = 0.000 Torsion = 92.752
===============================================================================
NSTEP = 11937 TIME(PS) = 5.968 TEMP(K) = 1042.07 PRESS = 0.0
Etot = -2354.7126 EKtot = 3771.9493 EPtot = -6126.6619
BOND = 958.4815 ANGLE = 1725.5483 DIHED = 1305.0448
1-4 NB = 428.7319 1-4 EEL = -4720.9001 VDWAALS = -670.3215
EELEC = 3408.4973 EGB = -8721.4561 RESTRAINT = 159.7119
EAMBER (non-restraint) = -6286.3738
------------------------------------------------------------------------------
NMR restraints: Bond = 66.935 Angle = 0.000 Torsion = 92.777
===============================================================================
NSTEP = 11938 TIME(PS) = 5.969 TEMP(K) =431725.07 PRESS = 0.0
Etot = NaN EKtot = 1562702.1324 EPtot = NaN
BOND = 977.0778 ANGLE = 1738.1119 DIHED = 1303.8070
1-4 NB = 431.4051 1-4 EEL = -4741.3679 VDWAALS = -668.7301
EELEC = 3396.5918 EGB = NaN RESTRAINT = 160.1925
EAMBER (non-restraint) = NaN
------------------------------------------------------------------------------
NMR restraints: Bond = 67.418 Angle = 0.000 Torsion = 92.775
===============================================================================
NSTEP = 11939 TIME(PS) = 5.969 TEMP(K) =********* PRESS = 0.0
Etot = NaN EKtot = 6236892.6366 EPtot = NaN
BOND = 1156.7229 ANGLE = 1812.9956 DIHED = 1306.7566
1-4 NB = 431.9067 1-4 EEL = -4729.3402 VDWAALS = -667.6747
EELEC = 3383.4688 EGB = NaN RESTRAINT = 160.3037
EAMBER (non-restraint) = NaN
------------------------------------------------------------------------------
Best regards
Peter
-----------------------------------------------------------------------
The AMBER Mail Reflector
To post, send mail to amber.scripps.edu
To unsubscribe, send "unsubscribe amber" (in the *body* of the email)
to majordomo.scripps.edu
Received on Wed Oct 01 2008 - 05:13:13 PDT
