AMBER Archive (2007)Subject: Re: AMBER: large RMS fluctuations in turning off VDW interactions by TI
From: David Mobley (dmobley_at_gmail.com)
Date: Mon Jul 02 2007 - 15:03:17 CDT
Holly,
I agree with Julien here, and also recommend the paper he mentioned,
as well as the Beutler and van Gunsteren paper (referenced therein) on
softcore potentials from about 1994. You might also have a look at
this preprint on a study on essentially this issue:
http://www.dillgroup.ucsf.edu/~dmobley/papers/steinbrecher.pdf (the
Beutler and van Gunsteren paper is also referenced there).
The next version of AMBER will have "soft core potentials"
implemented; these are designed to deal with this problem by removing
the singularity in dV/dlambda and reducing the numerical noise.
Best wishes,
David
On 7/1/07, Julien Michel <j.michel_at_soton.ac.uk> wrote:
> Hi Holly,
>
> Turning off LJ interactions can be problematic in FEP/TI simulations
> because the excluded volume of the atoms being turned off effectively
> only vanish in the very last steps of the coupling parameter. This is a
> well known issue that can be addressed by different techniques like
> retracting vanishing atoms in the VdW radius of other atoms, non linear
> scaling of the force field terms or a softcore (a modified LJ energy
> function). This paper may help you understand the issues and the solutions:
>
> "A Comparison of Non-Bonded Scaling Approaches for Free Energy
> Calculations" Pitera J.W.1; van Gunsteren W.F. Molecular
> Simulation,Volume 28, Number 1, 1 January 2002 , pp. 45-65(21)
>
> Julien Michel
>
> Holly Freedman wrote:
> > Dear AMBER support,
> > I have been struggling with a problem I've been having with a thermodynamic integration calculation.
> > I have a molecule with a little over 50 atoms, and I am perturbing a small part of it (O-COCH3) to (O-
> > CH3). According to advice from AMBER, I am dividng the calculation into four parts: Appearing and
> > disappearing atoms (ie the uncoupling of Van Der Waals interactions), and charging and uncharging
> > atoms which have appeared or disappeared. I am having trouble with the calculations where I
> > decouple Van Der Waals interactions because the RMS fluctuations in DV/DL which I am getting are
> > way too high (average over 3 kcal/mol). I am doing this using 9 values of clambda as described in
> > the manual.
> >
> > Here is my infile:
> > First equilibration:
> > &cntrl
> > imin = 0,
> > ntx = 1, irest = 0, ntr = 1,
> > ntpr = 100, ntwr = 100, ntwx = 5000, ntwe = 5000,
> > ntf = 2, ntb = 2, cut = 10,
> > nstlim = 500000, dt = .002, ntc = 2,
> > tempi = 0.0, temp0 = 298.0, ntt = 1,
> > tautp = 2.0, ntp = 1,
> > icfe = 1, clambda = 0.33787, klambda = 6,
> > &end
> > Restraining region
> > 10.0
> > RES 1 1
> > END
> > END
> >
> > Then:
> > &cntrl
> > imin = 0,
> > ntx = 5, irest = 1, ntr = 1, iwrap=1,
> > ntpr = 100, ntwr = 100, ntwx = 5000, ntwe = 5000,
> > ntf = 2, ntb = 2, cut = 10,
> > nstlim = 2500000, dt = .002, ntc = 2,
> > tempi = 298.0, temp0 = 298.0, ntt = 1,
> > tautp = 1.0, ntp = 1,
> > icfe = 1, clambda = 0.33787, klambda = 6,
> > &end
> > Restraining region
> > 10.0
> > RES 1 1
> > END
> > END
> >
> > My guess was that this was because the frcmod file I used for the dummy atoms wasn't consistent
> > with the force field for the analogous real atoms, but I can't see my mistake. Here is a my frcmod:
> >
> > remark goes here
> > MASS
> > D2 1.008
> > D4 12.01
> > D5 16.00 same as O
> >
> > BOND
> > D5-c 648.0 1.214 same as c-o
> > c-D4 328.3 1.508 same as c-c3
> > D4-D2 337.3 1.092 same as c3-hc
> >
> > ANGLE
> > D5-c-os 76.2 122.43 same as o-c-os
> > D5-c-D4 68.0 123.11 same as c3-c-o
> > D4-c-os 69.3 111.96 same as c3-c-os
> > c-D4-D2 47.2 109.68 same as c-c3-hc
> > D2-D4-D2 39.4 108.35 same as hc-c3-hc
> >
> > DIHE
> > X-c -D4-X 0 0.000 0.000 0.000 same as X -c -c3-X
> > X-c -c3-X 0 0.000 0.000 0.000 same as X -c -c3-X
> >
> > IMPROPER
> >
> > NONBON
> > D2 1.0000 0.0000
> > D4 1.0000 0.0000
> > D5 1.0000 0.0000 ATTN, need revision
> >
> >
> > I was wondering if anyone else has had a similar problem, and discovered what was going wrong?
> > I would really appreciate any helpful suggestions anyone has on this.
> >
> > Thanks in advance,
> > Holly Freedman
> > --
> > Department of Physics, University of Alberta
> > Edmonton CANADA
> >
> > -----------------------------------------------------------------------
> > The AMBER Mail Reflector
> > To post, send mail to amber_at_scripps.edu
> > To unsubscribe, send "unsubscribe amber" to majordomo_at_scripps.edu
> >
> >
>
> --
> Julien Michel
> http://www.julienmichel.net
> Dr J. W. Essex research group
> University of Southampton
> United Kingdom
>
>
>
> -----------------------------------------------------------------------
> The AMBER Mail Reflector
> To post, send mail to amber_at_scripps.edu
> To unsubscribe, send "unsubscribe amber" to majordomo_at_scripps.edu
>
-----------------------------------------------------------------------
The AMBER Mail Reflector
To post, send mail to amber_at_scripps.edu
To unsubscribe, send "unsubscribe amber" to majordomo_at_scripps.edu
|