Re: [AMBER] Thermodyncamic Integration (TI)

From: David A Case <>
Date: Fri, 27 Jul 2018 15:04:51 -0400

On Fri, Jul 27, 2018, emanuele falbo wrote:
> Ox + e- → Red
> by means of the TI method. Amber works well (only in serial) when the
> solvent is not implemented explicitly. However, when I include the solvent
> and ions in the systems sander doesn't work- I noticed it's because the
> final and initial structure must have the same coordinates.

In TI in sander, you run two simulations in parallel: they use different
Hamiltonians but the same coordinates. So, typically, you would use the
same coordinate file (pointed to by the "-c" flag) for both groups.
For solvated systems, you have to make sure that the number of waters
and ions is the same at the two end points: you cannot, for example,
have different numbers of counterions for the two different charge

[I don't understand why Ross and Mike decided to give these files
different names in the tutorials. If you look, files like
asp_model_step1.rst and asph_model_step1.rst are identical. Because
they were using GB, leap made identical files. But the coordinates must
be the same in both files, and this is generally done by giving the same
file name for "-c" in both groups.]

> 1. How do I include the solvent in the two structure keeping the
> coordinates of the two systems equal? (Ox and Red differ only for the
> charges).

See above; with modern Amber, I would build the Ox (say) state and
solvate it; then use the "change" command in parmed to create a new
prmtop file with altered protein charges. Then you have two prmtop
files, and one coordinate file, which is just what you need.

> 2. Why doesn't SANDER work in parallel?

Without knowing anything other that "it didn't work", no one can
help very much. What is the exact error message?


AMBER mailing list
Received on Fri Jul 27 2018 - 12:30:02 PDT
Custom Search