Re: [AMBER] Res-Res energy shows different value with or without explicit solvent

From: Jason Swails <>
Date: Tue, 19 Nov 2013 21:43:20 -0500

On Tue, Nov 19, 2013 at 9:12 PM, kurisaki <> wrote:

> Thank you for your suggestion.
> I gave external dielectric constant value of 1, i.e., "extdiel = 1"
> to also ignore electrostatic term in GB...
> > Carlos
> I'm sorry to forget show the column information with data.
> resid1 ->resid2 |internal |vdw |eel |pol |sas
> -----------------------------------------------------------------------
> With solvent
> TDC 79-> 79 0.000 -1.567 -5.462 0.000 0.000
> Without solvent
> TDC 79-> 79 0.000 -1.346 -5.709 0.000 0.000

OK. In this case, the two numbers should be the same. Figuring out why
they're not will require a bit more information. At this point, I
suspect the problem may be the input structures.

How did you generate them? What format was the input structure to sander?
 Was it an ASCII trajectory using imin=5? Was it an ASCII restart file
using imin=1 and maxcyc=0? Is there a reason the coordinates should be

If you used an ASCII file---particularly an ASCII trajectory file as is
generated automatically by (and used as input by
your numbers have very low precision. I've seen a simple RMS fit with an
ASCII trajectory file give rise to energy differences this large for a
single structure. Do you still get these differences when you use a full
binary (NetCDF) format for the input coordinates at every stage?

Jason M. Swails
Rutgers University
Postdoctoral Researcher
AMBER mailing list
Received on Tue Nov 19 2013 - 19:00:02 PST
Custom Search