Re: [AMBER] Interpreting energy decomposition in Amber thermodynamic integration

From: David Case <>
Date: Sat, 21 Jan 2017 22:05:02 -0500

On Fri, Jan 20, 2017, Stefan Ivanov wrote:
> I'm transforming a template ligand bound to a protein to a modeled
> ligand bound to the same protein with 9 λ windows (for starters). Say I
> look at the computed per-residue energies at λ = 0.1 and at λ = 0.9. I
> want to know what the reported energies mean at the specified λ values
> and how I can compare them, say for residue X at λ = 0.1 and at λ = 0.9.

You ask "what the reported energies mean": they are the average values
of dV/dlambda at the given lambda point. Only the integral of dV/dlambda has
any usual "meaning": it is a free energy.
> As I said, I'm doing decomposition with idecomp = 1 and only the ligands
> have any significant internal energy contributions and I was wondering
> why that is.

Ah: the word "significant" is important here. You are changing the ligand, so
the protein energies will only change to the extent that the protein
configuration differs between the two ligands. Unless residue 129 (say) is in
a different environment with ligand A vs ligand B, it will not contribute
to the free energy change. (One reason that I am not much of fan of TI energy
decomposition: it often tells you no more than what you could have figured
out with a careful visual study of the protein binding pocket.)

I encourage you to experiment with some very simple TI transformations to
get a clearer understanding of how everything works.


AMBER mailing list
Received on Sat Jan 21 2017 - 19:30:02 PST
Custom Search