Re: [AMBER] TI input files in A9 tutorial and Amber12 manual

From: Hamed S. Hayatshahi <>
Date: Fri, 6 Sep 2013 10:52:32 -0700 (PDT)

Thanks for the very helpful explanation. Now I am particularly focusing on understanding the implementation of simulations in your 2011 "Soft-Core Potentials in Thermodynamic Integration:Comparing One- and Two-Step Transformations".
For example, in the one-step Mg+2 system, you had an ion-containing water box as V0 and the same water box without ion as V1; and both crgmask and scmask masks were ':Mg+' for V0 and ' ' for V1. This, both vdW and charge interactions for the ion are expected to change from Mg+ parameters to nothing. Am I understanding it correctly? And so the integration over DV/DL values over the 99 windows on this one step added up to ~414 (according to table 3), right?

Hamed S. Hayatshahi
PhD candidate in Medicinal Chemistry
University of Utah
+1 801-867-4501

 From: "" <>
To: Hamed S. Hayatshahi <>; AMBER Mailing List <>
Sent: Thursday, September 5, 2013 1:27 AM
Subject: Re: [AMBER] TI input files in A9 tutorial and Amber12 manual


> 1) No soft-core potentials with need to dummy atom; and changing the
> charge and vdw in different steps as discussed in part 4.1.1 of the
> amber12 manual.
> 2) Soft-core potential used only in changing the vdw; and changing the
> charge and vdw in different steps in which no dummy atom is needed; as
> discussed in tutorial A9.
> 3) Soft-core potential used in changing the charge and vdw simultaneously
> in a single step with no dummy atom needed; as discussed in part 4.1.2 of
> the amber12 manual . 

This is a correct summary. Approach 1 is still possible for backwards
compatibility reasons and for a subset of TI problems where the atom
number remains constant such as computing charge transfer reorganisation

Approach 2 has in my opinion mostly been replaced by the more efficient
#3, but also remains in the code because a) there are cases when #3 shows
convergence problems and b) we dont have a tutorial for #3 yet (mea culpa)

> This way, A9 tutorial seems to me like a transient intermediate method
> between the methods introduced in part 4.1.1 and 4.1.2 of the manual. This
> is because it changes the charge and vdw in different steps while using
> sof-core potential in just one step out of three. Furthermore, it seems

tutotial A9 corresponds to #2, for pure changes of charge, softcore
potentials (neither for vdW nor coulomb) are not necessary. The initial
implementation of softcore (#2) was only intended for one of the typically
three steps of a TI transformation.

> that amber12 doesn't like parts of A9 where ifsc=0 while the coordinate
> and mdin files are set to be different for two ends of a step (V0 and V1),
> as expected from the manual 4.1.1. (as Dr. Cheatham says, tutorial might
> be out of date for amber12)

Yes, this may be confusing in the manual, since it tries to describe three
types of calculations in the same chapter. ifsc=0 always implies same
number and order of atoms in V0 and V1 (and will fail otherwise).

> Also, better for both the manual and tutorial to be more clear about the
> difference between crgmask and scmask. Does scmask apply only when ifsc=1?
> Should them be the same when using the soft-core potential to
> simultaneously change the charge and vdw? 

crgmask strips some charges out of your system during startup. It often is
set equal to scmask or covers a whole molecule. The important thing is,
crgmask is only a (somewhat hackish) option for system preparation, scmask
actually changes the forcefield equation (scmask only is used when
ifsc=1). So any setting of crgmask could also be replaced by editing your
prmtop files beforehand.

Kind Regards,


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

AMBER mailing list
AMBER mailing list
Received on Fri Sep 06 2013 - 11:00:03 PDT
Custom Search