Re: FEP implementation in NAMD 2.6 vs. 2.7b1

From: Chris Harrison (charris5_at_gmail.com)
Date: Wed Jul 29 2009 - 06:11:33 CDT

Dan,

To add to Floris' comment, you are seeing the non-physical path resulting
from the softcore potential as described by Zacharias, in which
electrostatics and vdW are independently decoupled from the MD potential.
Please look at the decoupling schemes figure in the FEP section of the
manual. As you will see there is a type of minimum at lambda=0.5 where
electrostatics for both incoming and outgoing partitions are at a minimum;
and vdW for both partitions are at the lowest values possible for
*both*partitions
*at the same time* *(or lambda value)*. This "middle-point" corresponds to
the "inflection point" in your non-physical path. The rest should be fairly
obvious; please let us know if not. We'd be happy to try and help clarify.

WRT the difference between 2.6 and 2.7b1, I suggest, off the cuff, that this
is due to the lack of a softcore potential in 2.6 and that as you increase
the number of non-linear windows as you approach lambda values of 0 or 1, I
suspect your dG will approach 0 .... as it does in the case of softcore with
2.7b1's value of -0.002.

Chris

--
Chris Harrison, Ph.D.
Theoretical and Computational Biophysics Group
NIH Resource for Macromolecular Modeling and Bioinformatics
Beckman Institute for Advanced Science and Technology
University of Illinois, 405 N. Mathews Ave., Urbana, IL 61801
char_at_ks.uiuc.edu                            Voice: 217-244-1733
http://www.ks.uiuc.edu/~char               Fax: 217-244-6078
On Tue, Jul 28, 2009 at 4:50 PM, Dan Bolintineanu <
dan.bolintineanu_at_gmail.com> wrote:
> Hi all,
>
> I've been trying to do some FEP calculations using NAMD, and have been
> experimenting with the ethane-ethane zero-sum calculation in the NAMD
> FEP tutorial to get a feel for FEP calculations.  I found the results
> a bit surprising, and was hoping someone could shed some light on
> them.
>
> Using NAMD 2.7b1 with
> fepElecLambdaStart 0.5
> fepVdwLambdaEnd 1.0
> decouple on
> I obtain a satisfactory final result, about -0.002 kcal/mol.  However,
> the lambda vs. delta G plot looks nothing like the one in the
> tutorial; it decreases linearly to about -7 kcal/mol, after which it
> increases linearly to -0.002 kcal/mol.
>
> In NAMD 2.6, using the exact same input file and sampling strategy,
> but commenting out the three lines above, I obtain a final result of
> +0.2 kcal/mol, but a delta G plot that seems much more reasonable,
> very similar to the one in the tutorial (except for the endpoint).
>
> In both cases, I have used much higher sampling (larger # of windows)
> than indicated in the FEP tutorial.  I've attached my input file for
> the NAMD 2.7b1 run.
>
> My questions are:
> 1.)  Why does the NAMD 2.6 final result not coincide with the one in
> the FEP tutorial?
> 2.)  Why does the NAMD 2.7b1 lambda vs. dG curve look so strange, and
> is the result valid in spite of this?
>
> Thanks in advance.
>
> Dan Bolintineanu
>

This archive was generated by hypermail 2.1.6 : Wed Feb 29 2012 - 15:53:06 CST