[AMBER] Minimisation error and system blowup for a large protein using implicit solvent

From: Joseph Salmon <js13559.2013.my.bristol.ac.uk>
Date: Wed, 22 Feb 2017 14:54:03 +0000

Dear All,

I have produced .nc files after the heating phase on a large complex. The
result of loading the file is that all the atoms shoot of in different
directions (system blows up). The minimisations prior to the heating seem
to have not worked, with a LINMIN failure reported in the mdout file. An
example of the minimisation input is:

Initial minimization of hydrogens


imin=1, maxcyc=5000, ncyc=1000, ntmin=1,

ntpr=1, ntwr=1,

ntb=0, cut=9999.0,

igb=5, rgbmax=15.0,

ntr=1, restraint_wt=25.0, restraintmask='!.H=',



And heating input:

Heating system from 0 K to 300 K over 200 ps


imin=0, irest=0, ntx=1,

dt=0.001, ntc=1, ntf=1,

ntb=0, cut=9999.0,

igb=5, rgbmax=15.0,

ntt=3, gamma_ln=2.0, ig=-1,

ntpr=1, ntwx=1, ntwr=100, nstlim=100,

nscm=1000, ioutfm=1,

ntr=1, restraint_wt=5.0, restraintmask='.CA,N,C,O',




TYPE='TEMP0', ISTEP1=1, ISTEP2=100000,

VALUE1=0.0, VALUE2=300.0,


&wt TYPE='END' /

(Please note that here I used ntpr, ntwx and ntwr = 1 to better monitor
what the problem was)

The complex contains two non-standard ligands which are not covered by the
standard protein force fields and so on set of parameters were used from a
previous study and the other were generated using the ff14SB protein and
GAFF parameters. Simulations have been run on AMBER-16, using CUDA on gpu
and also on cpu using sander.

The heating has been attempted without the gradual increase addition also
with and without a restraint (DISANG file). The minimisation has been
tested with ncyc = 10, 20, 50, 1000. Changes to the ff14SB have also been
trialled, for example the ff14SBonlysc, ff12SB. Finally the ligands were
removed and a minimisation and heating was run on just the protein
component and this still blew up.

The LINMIN failure only occurs during the conjugate gradient minimisation.
During the steepest descent minimisation, there are no error message.

Strangely, during the minimisation, even with ntwr=1, and 1000 steps of
steepest descent the structure written to the rst7 file is the same as the
original structure. See the attached mdout file from the minimisation.

Any help with regards to how to stop it would be greatly appreciated.

AMBER mailing list

Received on Wed Feb 22 2017 - 07:00:02 PST
Custom Search