Fwd: vmd-l: Re: hBond colvars and patching

From: Francesco Pietra (chiendarret_at_gmail.com)
Date: Tue Mar 08 2011 - 00:34:36 CST

Sorry, I forgot the mailing list. Now added

---------- Forwarded message ----------
From: Francesco Pietra <chiendarret_at_gmail.com>
Date: Tue, Mar 8, 2011 at 7:29 AM
Subject: Re: vmd-l: Re: namd-l: hBond colvars and patching
To: Peter Freddolino <petefred_at_ks.uiuc.edu>

Hi Peter: thanks for the comments and suggestions. For my comment,
please see below.

On Tue, Mar 8, 2011 at 12:08 AM, Peter Freddolino <petefred_at_ks.uiuc.edu> wrote:
> On 03/07/2011 11:04 AM, Francesco Pietra wrote:
>> Hello Peter:
>> Thanks for answering. Please, see below.
>> On Mon, Mar 7, 2011 at 4:19 PM, Peter Freddolino <petefred_at_ks.uiuc.edu> wrote:
>>> Hi Francesco,
>>> On 03/07/2011 01:12 AM, Francesco Pietra wrote:
>>>> this led to the added proton being in the unfavorable position that I
>>>> described, although the psf/pdb could be minimized and heated.
>>> Could you elaborate on how unfavorable this position seems? What's wrong
>>> with it? Unless you have something truly pathological, a minimization
>>> should give you an appropriate orientation.
>> Attached please see a picture of a GLU-ASP couple. At OE2 (highlighted
>> in cyan) of GLU I have attached a second proton (syn to ASP), at 180
>> deg from the original one (anti to ASP). That should illustrate that
>> the original placement of the proton is unfavorable for H-bonding with
>> ASP. Minimization should not (I guess) place the proton syn to ASP
>> because the original anti position has no unfavorable interactions.
> psfgen is doing its job here: place all atoms according to the
> definitions in the topology file. Moving atoms around further due to
> energetic considerations is a job for minimization or dynamics. If
> you're planning on doing what people typically do with these structures
> (running MD) this should not present a problem. If you're going to do
> something like docking and want a relaxed conformation, you'll need to
> run a short equilibration anyway. Do note that you don't *know* that the
> hydrogen rotation gives you the lowest energy structure anyway; there
> may also be some rotation of one or both of the involved residues as well.

As an experimental organic chemist, I can't dismiss the information
from crystal data (with due regard to possible differences in the
solution structure) and patch clamping. Crystal data clearly show
perfect H-bonding distances. Phenomena occur at low pH values implying
a good deal of protonation of GLU as if it were an isolated residue.
But it is not isolated: facing ASP, or another GLU, it pKa increases
considerably, which is a fact ascertained beyond any doubt. As patch
clamping is carried out in solution, I believe it would be simulating
something else than natural without having those H-bonds properly

No matter what I do, be that pure MD simulation, biased MD simulation,
or docking simulation, I do not move further if the system is not
equilibrated - or amenable to equilibration - assuming that the
equilibrated ensemble is the closest to nature. My suspicion - at the
moment - is that the what I am using does not treat H-bonding
properly. I'll report to you the result of equilibration of my
ensemble, starting from what autopsf allows to do, i.e., with
remaining "syn" situations.

Thanks a lot for your intervention


>> I am now trying again to set up in the pdb file from autopsf what I
>> believe is the correct stereochemistry, to run autopsf again. Maybe I
>> did a mistake in a previous similar attempt. However, I would be not
>> too much surprised that the matter can be tackled differently. These
>> are problems that people should have encountered each other day.
> A much better solution, if you really must have a starting structure
> with those hydrogens oriented the way you think they should be, is just
> to modify the coordinates in the pdb, and leave your psf the way it is.
> You should be able to do this either using molefacture, or with a little
> tcl scripting by using the standalone code discussed at
> http://www.ks.uiuc.edu/Research/vmd/mailing_list/vmd-l/12285.html

Thanks for the suggestion. I never used molefacture in VMD, it is the
right time now to learn it. However, from what I did in between my
last mail and your present answer, I am unsure whether any adjustment
of stereochemistry will solve the issue. I suspect (though I hope to
be wrong) that top_all27_prot_lipid.rtf (edited for the problem of H-H
into TIP3) I used in autopsf has an idiosyncrasy for H-bonds, at least
when interatomic distances are short (though perfect for H-bonds). In
fact, when the stereochemistry is adjusted to "syn" by the protocol I
have shown, autopsf brings back to "anti" most situations, except
those for the longer H-bonds.
> Best,
> Peter
>> Thanks a lot
>> francesco
>>>> ************
>>>> What I hope is that there is a mistake in my procedure, and be
>>>> corrected about. Otherwise suggestions how to set correctly H-bonds in
>>>> VMD/NAMD along a different route. As I said, I came to NAMD with
>>>> correct PDB files - from REDUCE or other - as far as H-bonds are
>>>> concerned. However, files strictly respecting PDB rules and, in
>>>> addition, also with some non CHARMM naming, which I tried to correct.
>>>> I was unable to arrive at workable psf/pdb along this route. Finally,
>>>> I could also correct the autopsf.pdb by repositioning the proton in
>>>> between GLU and the acceptor, be that the conjugate base or Cl-, with
>>>> a graphic package. However, this also failed - in my hands - to arrive
>>>> at workable psf/pdb.
>>> You would again need to give more details on what you tried to do and
>>> what errors occurred in order to get help.
>>> Best,
>>> Peter

This archive was generated by hypermail 2.1.6 : Mon Dec 31 2012 - 23:19:54 CST