Re: [AMBER] single -step transformation TI calculation

From: <>
Date: Fri, 9 Dec 2011 15:37:08 -0500 (EST)


> Problem: changing REF to NEW (actually only one atom is changed in the
> process)

This is a molecule not bound to anything else in your system I assume? You
want to go from solvated REF to NEW?

> ifsc=1, crgmask=':REF',scmask=':REF',
> ...............
> ifsc=1, crgmask=':NEW',scmask=':NEW',

Two things:

If you set crgmask like this, you change uncharged REF to uncharged NEW.
This would be the second step of a three-step transformation. For a
complete single point transformation, leave the crgmask parameter out
completely. The main point is that appearing/disappearing atoms now *can*
have charges.

More important though:

REF and NEW will not 'see' each other in this setup. This simulation
removes one residue completely and replaces it by another. Unless REF and
NEW are bound to something else, they can move to any part of the system.
This is a recipe for disastrous convergence problems, like people see in
absolute binding energy TI calculations. I think it might be better to
make parts of REF and NEW common atoms, if they are similar in any way, or
to introduce restraints (but take them into account correctly for total
dG, check B. Roux papers on this, for example).

Kind Regards,


Dr. Thomas Steinbrecher
formerly at the
BioMaps Institute
Rutgers University
610 Taylor Rd.
Piscataway, NJ 08854

AMBER mailing list
Received on Fri Dec 09 2011 - 13:00:03 PST
Custom Search