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

From: Francesco Pietra (
Date: Fri Mar 11 2011 - 09:07:20 CST

Hello Peter:
Done as you suggested. Now the stereochemistry in the couples
ASP-patchedGLU, GLU-patchedASP, and CLA-patchedGLU is substantially
the same before and after unrestricted minimization in a POPC membrane
(0 K, ts=1, "rigidBonds water", gradient 5.0) and seem palatable. See
attached examples. I hope the house is in order now.

I found no suggestion from the archives to decide which is best
appropriate, "distance" or "hBond" colvars, to restrict distances from
CLA to its ligand atoms (from patchedGLU, LYS, and ARG). Could that be
suggested? In fact, if the are problems about that, they will only
emerge from long simulations. My aim is to prevent CLA escaping, while
allowing formation of the H-bond during equilibration. To this
concern, it seems to me better not to restrain the GLU-ASP distance or
hBond, and see what the ff feels. If it is suitable to the system, it
is a way to see if the H-bond distances seen in the crystal hold for
the solution state as well in this forced "constant low pH"

Thousand thanks

On Fri, Mar 11, 2011 at 5:22 AM, Peter Freddolino
<> wrote:
> Hi Francesco,
> Could you try building your structure with "Regenerate angles/dihedrals"
> (under the autopsf Options menu) selected? That should get rid of the
> odd location of those hydrogens; in some cases psfgen fails to properly
> add all angle terms that should be present during patching, which would
> lead to what you saw. Also, the TIP3 topology listed below appears
> incorrect, as it gets rid of *all* of the bonds. You only need to get
> rid of the H1-H2 bond, which means deleting the last two words of the
> BOND line (H1 H2) but uncommenting it.
> Best,
> Peter
> On 03/10/2011 10:45 AM, Francesco Pietra wrote:
>> Hi Peter:
>> With VMD 1.9beta1/Linux
>> (1) Open the pdb for protein inserted into POPC (1.5A empty space
>> between protein and POPC. The latter, preequilibrated was taken from
>> charmm archives. The protein had been equilibrated at 310K/1atm in
>> previous work with AMBER10, force field ff99SB.
>> (2) Autopsf with top_all27_prot_lipid.rtf, edited as follows
>>     RESI TIP3         0.000
>>     GROUP
>>     ATOM OH2  OT     -0.834
>>     ATOM H1   HT      0.417
>>     ATOM H2   HT      0.417
>>     !BOND OH2 H1 OH2 H2 H1 H2
>>     (2a) Load input files
>>      (2b) Guess and split chains ....
>> segments were correctly identified (P1 for chain A; O1 for CLA in
>> chain A; P2 for chain B; O2 for CLA in   chain B; P3 for chain C; O3
>> for CLA in chain C; follow O4, O5, etc for POPC and W1..)
>>      (2c) Create chains
>> whereby DISU patches were automatically and correctly applied (the
>> S...S distances were typical)
>>      (2d) Apply GLUP and ASPP patches
>> which went on correctly as far as OE2 protonation is concerned (some
>> protons directed toward the acceptor O, some other ones turned back by
>> 180 deg, as I described previously).
>>      (2e) Apply patches and finish
>> getting psf/pdb.
>> (3) Add solvation box with above psf/pdb; Rotate 10deg; Boundary 2.4;
>> Use molecule dimensions; Box padding xyz 15 for min and max, followed
>> by AddIons, salt NaCl, isotonic (0.15M), minimum dist from solute and
>> between ions 5A, getting final psf/pdb.
>> (4) Measuring center and min-max, with the algebraic sum of min-max
>> for x giving the x cell dimension, and so on for y,z.
>> (5) Graphically checking that the patch-added protons were still where desired.
>> (6) Minimization ts=1fs, no restraints on H, first with protein and
>> its three CLA restrained, then unrestrained, final gradient 3.0.
>> During minimization, the GLU CG-OE2 rotated so that HE2 resulted
>> closer to the acceptor ASP (actually I have only checked that for one
>> patched-GLU--ASP in all three subunits).
>> *****
>> This from my notes and fresh memory. Should any doubt arise, or any
>> modification suggested as to the procedure, or simply a recheck being
>> desirable, I am here to re-do everything from scratch.
>> Thank
>> francesco
>> On Thu, Mar 10, 2011 at 3:36 PM, Peter Freddolino
>> <> wrote:
>>> Hi Francesco,
>>> I think there may be something seriously wrong with this structure...
>>> how exactly did you generate it (specifically, the psf)? What topology
>>> file did you use, and what commands were involved? I do not think the
>>> proton should be able to reach that position if the topology is correct--90e6ba4fbfe095f97c049e36514a--

This archive was generated by hypermail 2.1.6 : Mon Dec 31 2012 - 23:19:55 CST