Re: AMBER: results from TI

From: Thomas Steinbrecher <>
Date: Mon, 22 Oct 2007 10:35:15 -0700 (PDT)

Hi Holly,

I took a look at your input files and it looks like the run is set up ok.
Some comments, though: If you set noshake mask, you should set ntf=1 to
evaluate all potentials (the completed softcore code will check for this,
the version you used was probably before that). Also, you dont need to set
noshakemask for an atom that is not in scmask (but it shouldnt hurt you
much). You could try a short test run if you get about the same dvdl
result when removing noshakemask in the charging steps (but keeping it
set in the ifsc=1 vdw change step) with ntf set to one.

Do your trajectories look ok, with enough sampling of your compound?
Especially, do the softcore atoms occupy reasonable positions?

How do you results compare to experimental data? MM-PBSA can certainly be
off by 5 kcal or more especially in charged systems, so the difference
doesnt suprise me (of course, TI can be pretty far off too if its
unconverged etc and I agree that an OH group should be expected to
be better solvated than a MetO group.).

How do the free energy curves look like? A three point integration is the
bare minimum for a sizable charge change and nine points can easily be too
little for a vdw change. Check if your curves look smooth, otherwise the
integration can introduce some errors. Add some lambda-points if unsure.

You wrote that the magnitude of deltaG in some cases suprised you. Be sure
to check the output file for the dvdl-breakdown. Sometimes e.g. 1-4
interactions can add a lot of free energy change that cancels out in
another step. Take a look how much of the deltaG actually comes from
nonbonded interactions.

The files you sent are only for the charge removal of the H, which is the
step where the least can go wrong. You can sent me the other input files,
I'll take a look if you want.

Also, (I know this is rather trivial, but I did it wrong more than once)
are you sure you applied the right signs to all your free energy terms?



Dr. Thomas Steinbrecher
The Scripps Research Institute
10550 N. Torrey Pines Rd.
San Diego CA 92037, USA
The AMBER Mail Reflector
To post, send mail to
To unsubscribe, send "unsubscribe amber" to
Received on Wed Oct 24 2007 - 06:07:34 PDT
Custom Search