VMD-L Mailing List

From: Robin Betz (robin_at_robinbetz.com)
Date: Wed Oct 07 2020 - 18:26:41 CDT

Hi Harper,

I agree that using a CHARMM "patch" is the best way to define the
isopeptide bond, which actually ends up being technically simpler than the
AMBER method outlined in the tutorial.
Regardless, you still have to find parameters for the bond somewhere.

The way I like to do it is to draw out both residues, labelled with CHARMM
atom types (this will also help you learn the common atom types pretty
quickly).
During my PhD I found drawing out molecules on paper like this was always
worth the time it took, in terms of helping me learn the forcefields,
building my chemical intuition, and catching subtle errors.
Combining this with reading the forcefield topology and parameter files
directly will really make you learn how a forcefield works. They're just
text, easy to understand, and well commented.

You can find the atom types by visualizing the protein sans isopeptide bond
in VMD, or by reading top_all36_prot.rtf (part of the CHARMM36 forcefield)
which has little ASCII art drawings with atom names, which you can match to
types
based on the information provided (start by grepping for "RESI GLU"). Text
descriptions of atom types are found in par_all36m_prot.prm. I don't
usually draw the hydrogens although in this case I would on the atoms that
will be bonded.

Then, you can walk through what a patch would entail. You'll delete the
hydrogen atoms, add a bond, and then define parameters. Think about which
bonds, angles, and dihedral parameters are now present that weren't before.
You can grep around in par_all36m_prot.prm to see if any of them are
already defined. Don't forget that angles and dihedrals can be listed
forwards and backwards, for example an angle A-B-C could have a valid angle
parameter for C-B-A.

It looks like you can borrow a lot from the DKAM patch, as the atom type of
the carboxylate oxygens for both GLU and ASP is OC, "carboxylate oxygen".
Extra fortunately, the partial charges are
the same too, -0.76.

CHARMM doesn't care what residues are, only atom types and partial charges
matter in terms of what calculations happen in simulation. Residues are
really only helpful in helping humans work with and build these systems :)
It might be as simple as changing the atom and residue names in the patch!
Really do draw it out on paper and see though. Since you're learning, I'd
suggest drawing out LYS and ASP, then applying the DKAM patch on top of
your drawing, then doing the same for your LYS - GLU system and going from
there.

I agree that this stuff is really hard to learn. Feel free to email me your
drawings and progress off-list and I can help more.

Hope this helps,
Robin

On Wed, Oct 7, 2020 at 3:32 PM Axel Kohlmeyer <akohlmey_at_gmail.com> wrote:

>
>
> On Wed, Oct 7, 2020 at 4:53 PM Smith, Harper E. <
> smith.12510_at_buckeyemail.osu.edu> wrote:
>
>> Hi Josh,
>>
> Is there a more standard way to proceed for someone with weak chemical
>> intuition?
>>
>
> In my personal opinion, the best approach to this is to find an
> experienced collaborator. It used to be the common way back in the times
> when tools were less easy to use and information on how to use them was not
> as readily accessible to anyone as now.
>
> While it is much easier to get started now and do good work on
> straightforward projects, anything more complex has become more risky,
> since people don't go through an "apprenticeship" period anymore where they
> learn the "tricks of the trade" and about the non-obvious sources of errors
> that are known to experienced practitioners, but rarely put in writing
> (mostly because they are not really of the "do this, not that" category and
> thus not simple to write about).
>
> Axel.
>
>
>> Best,
>> Harper Smith
>>
>
>
> --
> Dr. Axel Kohlmeyer akohlmey_at_gmail.com http://goo.gl/1wk0
> College of Science & Technology, Temple University, Philadelphia PA, USA
> International Centre for Theoretical Physics, Trieste. Italy.
>