On Thu, Feb 22, 2024, Helmut Carter via AMBER wrote:
>Subject: Thermodynamic Integration of Loosening Restraints
>
>
>I would like to determine the free energy change associated with loosening
>restraints on a small molecule, going from 100 kcal/mol*A^2 to 0. Because
>this alchemical transformation does not involve a change in the molecule
>topology, I do not see a straightforward way of doing it in AMBER/pmemd.
Some coding would be needed here. Remember that TI basically computes the
average of dV/dlambda and various intermediate lambda values. Restraints
are computed in ene.F90 (subroutine xconst) in sander. You would need to
modify this code to make the restraints lambda-dependent, then compute the
average value over a simulation. For your case, you don't need to specify
icfe, since there is not need to consider two different prmtop files. You
would just run a straightforward simulation and print out (or average on the
fly) the required value.
(Same idea for pmemd, but I'd try in sander first. If you want GPU
acceleration, you would need to code a GPU kernel for that. So, consider
exploring the idea on the CPU version first.)
A caveat: I've never done this, and I might well be missing something.
Certainly, you would need to be careful if there are large
conformational changes upon relaxing the restraints -- e.g. if the small
molecule dissocates from the protein when the restraints go to zero.
Also: are you sure that what you describe is what you really want to do? The
"usual" procedure is to decouple the small molecule from the rest of the
system while restraints are still in place. After that, you can compute the
free energy of releasing the restraints analytically, since the small
molecule no longer interacts with the rest of the system. But your approach
might have some advantages; I'm just suggesting that you think carefully
about what you will do with your result once you obtain it.
...good luck....dac
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Fri Feb 23 2024 - 11:30:03 PST
