AMBER Archive (2009)

Subject: Re: [AMBER] g(r) extrapolation

From: Thomas Cheatham III (
Date: Wed Feb 18 2009 - 11:15:36 CST

> I obtained a couple of AMBER trajectories and I calculated the RDF
> with ptraj, and then calculated the structure factor by Fourier
> transformation with a in-house code.
> The fit with experimental data is very good at intramolecular
> distances (high k values), but gets worser at high distance..I guess
> that there could be some problems with the finite-size effects, since
> ptraj RDF are (obviously..) calculated within the hypothesis of
> minimum image convention, so that the maximum r value is half the box
> dimension.
> Does any body have suggestions on how to estimate the g(r) outside
> the box and/or suggest me some papers on this issue?

The ptraj code could be modified to include the longer range g(r)
information, however it is not clear if this would be that useful since it
would likely be biased by finite-size effects (essentially, continuously
running into copies of the solute in an ordered array rather than random

Two thing's I would try:

(1) run with a bigger box

(2) extrapolate the RDF to unity (1.0) from 1/2 the box size on... This
will effectively omit the long range components but not be biased by the
artificial drop of g(r) as you approach sizes greater than 1/2 the box
size due to the minimum image.


AMBER mailing list