Re: Question on making 'reversible' solvated simulation

From: Randy J. Zauhar (r.zauhar_at_usciences.edu)
Date: Mon Feb 19 2018 - 12:42:36 CST

Hi Brian,

Actually my plan was to send the active-site water cluster to vacuum at same time the ligand is being created (vice versa for the reverse simulation). That way the active site is blocked to entrance of additional waters, and there is a simultaneous estimate of site de-solvation. At least that is my fantasy, ;-)

I was off-base in my earlier message to Jerome, in fact the big energies I saw were due to something being horribly f-cked up after I tried to merge the ligand back into my apo simulation (I used the Merge tool in VMD). Not sure what happened, but it is clearly broken and I need to rebuild that.

Thanks!

Randy

On 19Feb, 2018, at 12:00 PM, Brian Radak <brian.radak_at_gmail.com<mailto:brian.radak_at_gmail.com>> wrote:

I've never dealt with displacing water molecules before. In addition to the trickiness of the chemical potential of water, choosing your restraints would likely be really tricky. You could also "displace" the water molecules alchemically by having them on top of the ligand in one endpoint and then move them into the bulk at the other, but you'd still have to no exactly how many water molecules you are talking about (maybe try multiple counts?).

This might also muck up performance a bit because the alchemical region is spread out, but it would simulate the bound/bulk transition explicitly.

On Sat, Feb 17, 2018 at 12:20 PM, Randy J. Zauhar <r.zauhar_at_usciences.edu<mailto:r.zauhar_at_usciences.edu>> wrote:
Hi Brian,

Is there any reason this would not work acceptably?

1) Equilibrate apo protein in solvent, active site is restrained in ligand-bound conformation

2) At end of equilibration, identify waters in active site by overlap with ligand atoms when bound.

3) Restrain active site water positions and re-equiibrate

4) Annihilate waters while creating ligand, and vice-versa to check reversibility. If I am correct I just need a merged PSF with both active site waters and ligand, and tag one group for creation, the other for annihilation.

I have not thought about the actual thermodynamic cycle that implies - Probably would need to do the same thing in bulk solvent (i.e. extract the active site waters and solvate them in a box as part of the annihilation/creation of ligand in bulk) ??

Thanks!

Randy

On 12Feb, 2018, at 11:00 AM, Brian Radak <brian.radak_at_gmail.com<mailto:brian.radak_at_gmail.com><mailto:brian.radak_at_gmail.com<mailto:brian.radak_at_gmail.com>>> wrote:

A rather involved, but essentially equivalent route that could be pursued in NAMD is to apply an orthogonal bias (or series of biases) on the water molecules. The bias energy could then be removed in a reweighting step (this goes beyond the normal procedure implemented in ParseFEP, for example, so you should probably be comfortable writing your own WHAM code in order to attempt this).

I've never tried this, but I believe it could be implemented with boundary forces (tclBC)? You might profitably reduce the overall number of simulations by using an adaptive method like ABF or metadynamics to get a global bias (as opposed to a local bias like a harmonic restraint). Again, this is really complicated.

Accelerated MD (or Gaussian accelerated MD) or even simulated tempering might also work. We're actively trying to couple alchemical calculations with REST2 for essentially this purpose, but that will not be released until NAMD2.14 at the earliest (yes, that far ahead).

HTH,
Brian

On Sun, Feb 11, 2018 at 1:52 PM, Randy J. Zauhar <r.zauhar_at_usciences.edu<mailto:r.zauhar_at_usciences.edu><mailto:r.zauhar_at_usciences.edu<mailto:r.zauhar_at_usciences.edu>>> wrote:
Jerome, thanks - I see the issue of site de-solvation not being properly included. ;-(

One thing I will check is my protein preparation - it is possible that I removed crystallographic active site waters, and the ones that crept in are in fact occupying similar positions. In that case I could simply include the xtal waters. That would at least improve things.

Another would be to retain the ‘solvent wall’, and introduce some active site waters in place of the ligand. Could then do a separate simulation to annihilate the active site waters, which would also provide some measure of de-solvation by the ligand.

Ultimately we want to compare ligands for the same target, and the desolvation of the site should be similar for each.

Randy

P.S. I will take a look at the work by Yang, but do not contemplate moving to CHARMm over this!

On 11Feb, 2018, at 5:56 AM, Jérôme Hénin <jerome.henin_at_ibpc.fr<mailto:jerome.henin_at_ibpc.fr><mailto:jerome.henin_at_ibpc.fr<mailto:jerome.henin_at_ibpc.fr>><mailto:jerome.henin_at_ibpc.fr<mailto:jerome.henin_at_ibpc.fr><mailto:jerome.henin_at_ibpc.fr<mailto:jerome.henin_at_ibpc.fr>>>> wrote:

Hi Randy,

While adding a barrier to entry for water molecules may well make the simulation more reversible, it will also make it biased and give an overestimated binding affinity, by killing the binding site desolvation contribution.

It's a tough problem, but ideally this solvation/desolvation process needs to be sampled.

Improving the sampling of that kind of "orthogonal" process was the focus of some work by Wei Yang:
http://www.pnas.org/content/105/51/20227.full
https://pubs.acs.org/doi/abs/10.1021/ct200726v

But that's only implemented in CHARMM afaik. The theory is somewhat involved.

So I'm sorry to say, if it's a common problem, there is no universal, simple and consensual approach to it.

Best,
Jerome

On 10 February 2018 at 19:15, Randy J. Zauhar <r.zauhar_at_usciences.edu<mailto:r.zauhar_at_usciences.edu><mailto:r.zauhar_at_usciences.edu<mailto:r.zauhar_at_usciences.edu>><mailto:r.zauhar_at_usciences.edu<mailto:r.zauhar_at_usciences.edu><mailto:r.zauhar_at_usciences.edu<mailto:r.zauhar_at_usciences.edu>>>> wrote:
Hi,

I am running a ligand annihilation in a solvated system, and I would like to make it truly reversible if possible.

Issue is that a couple water molecules sneak into the active site when the ligand disappears. Then when I reverse and start creating the ligand, they are trapped.

A simple thing I will try is to constraint the water molecules close to the active site, hopefully creating a barrier to entrance.

However. this has to be a pretty common problem, is there a recommended approach to address it?

Thanks!
Randy

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<tel:%28215%29596-8691><tel:%28215%29596-8691>
FAX: (215)596-8543<tel:%28215%29596-8543><tel:%28215%29596-8543>
E-mail: r.zauhar_at_usciences.edu<mailto:r.zauhar_at_usciences.edu><mailto:r.zauhar_at_usciences.edu<mailto:r.zauhar_at_usciences.edu>><mailto:r.zauhar_at_usciences.edu<mailto:r.zauhar_at_usciences.edu><mailto:r.zauhar_at_usciences.edu<mailto:r.zauhar_at_usciences.edu>>><mailto:r.zauhar_at_usciences.edu<mailto:r.zauhar_at_usciences.edu><mailto:r.zauhar_at_usciences.edu<mailto:r.zauhar_at_usciences.edu>><mailto:r.zauhar_at_usciences.edu<mailto:r.zauhar_at_usciences.edu><mailto:r.zauhar_at_usciences.edu<mailto:r.zauhar_at_usciences.edu>>>>

“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"

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<tel:%28215%29596-8691><tel:%28215%29596-8691>
FAX: (215)596-8543<tel:%28215%29596-8543><tel:%28215%29596-8543>
E-mail: r.zauhar_at_usciences.edu<mailto:r.zauhar_at_usciences.edu><mailto:r.zauhar_at_usciences.edu<mailto:r.zauhar_at_usciences.edu>><mailto:r.zauhar_at_usciences.edu<mailto:r.zauhar_at_usciences.edu><mailto:r.zauhar_at_usciences.edu<mailto:r.zauhar_at_usciences.edu>>>

“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"

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<tel:%28215%29596-8691>
FAX: (215)596-8543<tel:%28215%29596-8543>
E-mail: r.zauhar_at_usciences.edu<mailto:r.zauhar_at_usciences.edu><mailto:r.zauhar_at_usciences.edu<mailto:r.zauhar_at_usciences.edu>>

“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"

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: r.zauhar_at_usciences.edu<mailto:r.zauhar_at_usciences.edu>

“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 : Wed Nov 20 2019 - 23:19:41 CST