Re: [AMBER] parameters in mm/pbsa calculations using python and perl versions

From: Jason Swails <>
Date: Mon, 12 Dec 2011 16:21:54 -0500

On Mon, Dec 12, 2011 at 9:00 AM, Qinghua Liao <>wrote:

> Dear amber users,
> In python pbsa, the default option of* inp* is 1 (it is not clear whether 2
> is available in the manual), and the default *cavity_surften* and *
> cavity_offset* are 0.00542 and -1.008 respectively. However, in the perl
> version, the explanations from the manual are different:
> If INP = 1 and RADIOPT = 0 (default), use SURFTEN/SURFOFF parameters that
> fit with the radii from the prmtop file, e.g., use SURFTEN:
> 0.00542; SURFOFF: 0.92 for PARSE radii. If INP = 2 and RADIOPT = 1, these
> two lines can be removed, i.e., use the default values set in pbsa for this
> nonpolar solvation model. Otherwise, set these to the following: SURFTEN:
> 0.04356; OFFSET: -1.008
> So why is the default value of cavity_offset -1.008, but not 0.92 in the
> python version? Is inp of 2 available in python version? Another question,
> will the results be the same calculated by python and perl versions using
> the same parameters? Does someone test it?

The defaults in the Python version were adjusted to match the defaults used
in PBSA itself, rather than the defaults used in This is the
reason for the difference. Furthermore, exposes all of the
variables from PBSA that are present in the NAB interface (see the NAB
documentation regarding inp) or sander (see the Amber 11 source code). If
INP==2 is available there, it should be available in, too.

When we first wrote the Python version, we carefully tested the results
between the Python and Perl versions to see that they matched to reasonable
certainty. However, a couple things will always cause differences. The
use of (very) limited precision in ASCII trajectory/coordinate files makes
the results change with molecule orientation (since internal coordinates
change when limited-precision cartesians are arbitrarily rotated and/or

Within the tolerances of these effects, the two versions give the same
answers as long as you provide exactly the same parameters.


Jason M. Swails
Quantum Theory Project,
University of Florida
Ph.D. Candidate
AMBER mailing list
Received on Mon Dec 12 2011 - 13:30:03 PST
Custom Search