Re: [AMBER] Soft-core alchemical transformation in TI

From: <>
Date: Tue, 24 Jul 2018 08:49:47 -0400

Hi Simon,
        Here are our answers:

>So, I only have to set the timask and scmask on the atoms as in the tutorial? All atoms present in timask while not in scmask
>will be transformed without soft-core potential automatically?

        Yes. All atoms in timask but not in scmask will be treated as common atoms, provided that each common atom can find its partner atom and all common atom pairs have the same coordinates (up to 0.1 A tolerance).
>Second, I already simulated methanol solvated without the topology of ethanol. If I want to keep the coordinate and *velocity*
>for a new TI, how should I prepare the files? The tutorial is preparing benzene and phenol from a pdb file.
>Can I skip it by giving a rst file with coordinate and velocity information?

        You could do that but not just directly using a methanol restart file. The restart file only contains methanol coordinates. You need a restart files containing both methanol and ethanol coordinates--although they could be the same. You might use the tiMerge utility (manual page 275) in the ParmED module (manual page 252) to merge a methanol restart file w/ an ethanol restart file. It takes some time to learn how to use it but it will be very useful if you want to TI/FEP/MBAR calculations often.

Taisung & Dave

-----Original Message-----
From: Simon Kit Sang Chu <>
Sent: Monday, July 23, 2018 10:44 PM
To: AMBER Mailing List <>
Subject: Re: [AMBER] Soft-core alchemical transformation in TI

Dear David,

Thanks for the info. I am also looked into the same tutorial. But I am still confused to add soft-core potential atoms.

According to AMBER16 manual, it seems that I do not have to add dummy atoms for my case.

> Note that a slightly different setup is required for using soft core
> potentials compared to older TI- implementations. Specifically, the
> difference is that to add or remove atoms without soft core
> potentials, they are transformed into interactionless dummy particles,
> so both end state prmtop files have the same number of atoms. When
> using soft core potentials instead, no dummy atoms are needed and the
> end states should be built without them.


David Cerutti <> 於 2018年7月24日週二 上午12:57寫道:

> There are two ways to do it:
> 1.) Map the extra dummy atoms of ethanol to methanol: in this way,
> each atom has its exact mapped partner in methanol and ethanol, and
> they will have exactly the same coordinates. The masses of atoms do
> not affect the binding free energy so you don’t need to worry.
> 2.) Treat different atoms as “softcore regions.” In this way, the
> different atoms will move independently.
> The second way is preferred, as the result will usually be much more
> stable.
> You can find the parameter settings in the manual (p426) and a
> step-by-step example in
> Best of luck!
> Dave and Taisung
> On Mon, Jul 23, 2018 at 11:12 AM Simon Kit Sang Chu <
> wrote:
> > Hi everyone,
> >
> > I am planning to transform a methanol into an ethanol. I did a
> > simulation of pure methanol solvated in water. I want to keep all
> > the coordinates
> and
> > velocities, including common hydrogens, in the restart in amber
> > format. I have two concerns right now.
> >
> > First, to transform from methanol to ethanol, I have to cut a
> > hydrogen
> out
> > from methanol and append a CH3- from the broken end. I briefly look
> > into AMBER manual and dummy atoms are necessary for pmemd
> > preparation. Atoms transformed must also have the same masses so I
> > cannot transform the hydrogen truncated into the new carbon. In that
> > case, how should the coordinate files be written? Can I skip
> > creating a pdb with a crd and include the hydrogens?
> >
> > Second, I am new to AMBER and I am not sure if it is using dual topology.
> > If so, the CH3- will still be bonded with methanol while having no
> > LJ and electrostatics. However, will the thermal motion created by
> > the
> thermostat
> > alter the motion of methanol even at lambda = 0? If I am
> > transforming a methanol to a hexanol, the methanol motion would
> > largely be random due to the higher momentum of the "invisible" (CH2)4-CH3 tail.
> >
> > I appreciate any advice. Sorry for the long mail!
> >
> > Regards,
> > Simon
> > _______________________________________________
> > AMBER mailing list
> >
> >
> >
> _______________________________________________
> AMBER mailing list
AMBER mailing list

AMBER mailing list
Received on Tue Jul 24 2018 - 06:00:07 PDT
Custom Search