Re: Question on restraint behavior when part of system decouples

From: Jérôme Hénin (
Date: Mon Jan 29 2018 - 16:20:22 CST

In your cycle you remove the T and O restraints in solvent. That seems
perfectly fine to me, but if they are removed one step earlier in vacuum,
the effect is really the same. The steps in bulk do not need to mirror
those in the binding site, the only requirement is to close the cycle. Then
it's whatever floats your boat.


On 29 January 2018 at 22:58, Randy J. Zauhar <> wrote:

> Hi Jerome,
> I may be pondering this too much, but to me something feels fundamentally
> inconsistent - the constraints to the ligand in vacuum are different in the
> bulk VS the complex simulations, if following the tutorial protocol.
> I may be beating a dead horse over a small effect, but I think this cycle
> is better:
> In the above, I have the ligand in vacuum with translational (T),
> orientation (O) and conformation (C) restraints; the additional X* in the
> constrained complex represents any additional restraints that need to be
> imposed (e.g. in a recent simulation I had to include a restraint to
> prevent a salt bridge from melting and crashing into the ligand as it
> decoupled).
> This cycle sums to zero, but if I have ONLY conformational constraint in
> ligand bulk simulation, it does not!
> (The same constraints can be imposed on the bulk simulation by (e.g.)
> adding a dummy atom for the translational restraint. )
> The total free energy expression for this cycle is
> deltaG(bind) = (deltaG(cmplx*) - deltaG(lig*)) + (deltaG(cmplx) -
> deltaG(lig))
> Here is some nice intuitive physics - the constraints for the solvated
> ligand in bulk will be ‘active’, since there is nothing BUT the constraints
> to hold the ligand in place, and deltaG(lig) will be a significant negative
> number.
> In contrast, the constraints in the complex will be comparatively ‘dead’,
> as the surrounding active site is already embracing the ligand, and
> deltaG(cmplx) will be a negative number, but of smaller magnitude.
> So the ‘constraint’ component of the binding free energy, (deltaG(cmplx)
> - deltaG(lig)) > 0 , which makes perfect sense - the entropy has decreased
> upon binding of the ligand, and that decrease is seen in all three
> components (translation, orientation, conformation).
> Let me know what you think,
> Randy
> On 29Jan, 2018, at 2:52 PM, Jérôme Hénin <<mailto:j
>>> wrote:
> The protocol I have in mind is the one shown in the NAMD free energy
> tutorial, “Protein:ligand standard binding free energies…”, especially Fig.
> 1 and associated narrative (Paragraphs numbered 1-5, copied below). The
> idea is that only the conformation of the ligand is ‘pinned’ in the bulk
> simulation, while in the complex, ligand position, orientation AND
> conformation are restrained. I would think that the three types of
> constraint ‘capture’ three important components of small molecule binding
> in terms of entropy loss, and don’t understand why only one of these
> contributions is ‘picked out’ in the bulk simulation.
> Because that's the only one that will make a difference. Translation and
> rotation are just symmetry operations in the isotropic bulk.
> Put simply, why not just annihilate the ligand in solution with no
> constraints?
> That would be an option: but then you'd have to release the conformation
> restraints in vacuum to get the free energy contributions. One reason to
> avoid this is, if you have a flexible ligand like a peptide, the decoupling
> calculation in bulk will converge faster under restraints. Then the
> assumption is that the favored conformations in solution are not too
> different than those in the binding site, so releasing the restraints there
> is not too drastic.
> And by the way, the tutorial implies that the purpose of the restraints is
> to avoid the ‘wandering ligand’ problem, where as the ligand decouples it
> starts moving around and gets tangled up with the protein, leading to huge
> energies and poor accuracy, a very real problem! - but I don’t see where
> it discusses entropy?
> Well, not explicitly perhaps. It's included implicitly if the restraint
> free energy calculations.
> Best,
> Jerome
> Randy J. Zauhar, PhD
> Prof. of Biochemistry
> Dept. of Chemistry & Biochemistry
> University of the Sciences in Philadelphia
> 600 S. 43rd Street
> Philadelphia, PA 19104
> Phone: (215)596-8691
> FAX: (215)596-8543
> E-mail:<>
> “Yeah the night is gonna fall, and the vultures will surround you /
> And when you’re lookin’ in the mirror what you see is gon’ astound you"
> — Death Cab for Cutie, “Monday Morning"

This archive was generated by hypermail 2.1.6 : Tue Dec 31 2019 - 23:19:38 CST