RE: AMBER: nonpolar solvation energy

From: Ray Luo <>
Date: Wed, 31 Jan 2007 11:37:34 -0800

You can definitely compute solute-solvent vdw interactions by running MD in
TIP3P/PME. This is exactly how the nonpolar solvent in amber9 was calibrated
in the first place. However, you need:

1) Run MD for each MMPBSA snapshot with all solute atoms restrained.

2) Make sure all above runs are fully equilibrated. You can check the
quality of equilibration to looking at the running average of vdw energy for
each of the runs.

3) Once you are sure the MD runs are fully equilibrated, you need to use
ptraj to generate at least 100 inpcrd files (fully uncorrelated, i.e. at
least 5ps apart) in the equilibrated portion of the trajectory for each
MMPBSA snapshot. You need three set of inpcrd files for each MMPBSA
snapshot, one for solute/solvent, one for solute only, and one for solvent

4) Then you compute mean vdw energies for the three sets for each MMPBSA
snapshot. AND you have to change the sander program to make sure that you
only compute the attractive portion of the vdw energy. This attractive
portion of the vdw energy is what you need in MMBPSA using the new nonpolar
solvent model.

5) Finally use <U(solut_solv)> = <U(total) - U(solut) - U(solv)> to compute
the needed solute/solvent vdw energy for each MMPBSA snapshot.

Finally you can go back to your usual MMDPBSA calculation. So the bottom
line is that it is extremely involving and I strongly recommend that you
upgrade to amber9 to use the new PBSA for this.

All the best,

Ray Luo, Ph.D.
Dept Molecular Biology & Biochemistry
University of California, Irvine, CA 92697
USPS: PO Box 3900
Phone: (949) 824-9528

hi all

i have a question regarding MMPBSA in amber8
since the program doesnt calculate the solute-solvent
vdW interaction or the dispersion contribution as
regard to the nonpolar solvation energy, is there any
way that i could do it manually? my knowledge is quite
limited, can i do it by running sander in explicit
solvent for each extracted snapshot and just take the
vdW contribution? but the values would be so huge, is
that correct? because it would involve solvent-solvent
interaction as well (is it?) and how many steps should
i run? is it enough if just 1 step? i really need
someone to guide me on this

i know that in amber9, the dispersion contribution has
been implemented in PBSA, thats really neat but i dont
have amber9.

if there are other suggestions on how i should extract
the solute-solvent vdW contributions, please shout
out, i would definitely definitely appreciate it very
very much!

thanks all

The AMBER Mail Reflector
To post, send mail to
To unsubscribe, send "unsubscribe amber" to
Received on Sun Feb 04 2007 - 06:07:13 PST
Custom Search