From: mariano spivak (
Date: Tue Jul 09 2019 - 17:56:47 CDT

Hi again,

I tested the set up and the calculations and this is my take on your issue.

The ligand that you want to parametrize is highly charged (-4), which I
believe is causing the water interaction issues. Some of the files are
completely wrong given the clashes between the probe water and phophate
(DON-H2 for example) and I don't think you can use those results. On the
other hand, files where the probe water is repelled by the ligand are most
likely caused by the large electrostatic interaction between the phosphate
groups. I have not tried, but it is possible that using a "better" QM
method for these water interaction files would solve this problem. For
example, in the problematic gau file (DON-H3), replace HF by B3LYP.

If that doesn't work, I would suggest you try the RESP fitting method and
compare with the charge values of the standard chemical groups in CGenFF.

As a side point, I played around with the ligand and found a lower energy
structure than the one you are using in the optimized pdb. The "new"
minimum is actually not entirely linear but forming a intramolecular
hydrogen bond between the OH and the PO4, which results in a 7 member ring
connecting atoms 12 (H) and 7 (O). If you are curious you might want to
test this.


Mariano Spivak, Ph.D.

Theoretical and Computational Biophysics Group

Beckman Institute, University of Illinois.

On Sun, Jul 7, 2019 at 11:18 AM WT Ren <> wrote:

> Dear Mariano:
> Thanks very much for your reply. I uploaded the files to google drive.
> The initial PDB structure was constructed by Molefacture.
> Now I'm confused about which QM log files should be used and how to
> decide the proper scale parameter and the weight of distance/dipole.
> Thanks in advance!
> Best regards
> Weitong Ren
> On Mon, Jul 1, 2019 at 1:01 AM mariano spivak <> wrote:
>> Hi,
>> I can check your files to see what is the problem. You can send me the
>> files in a separate email.
>> It sounds like you are including water molecules from your pdb/xray
>> structure in the FFTK framework, which is not needed.
>> Best
>> Mariano Spivak, Ph.D.
>> Theoretical and Computational Biophysics Group
>> Beckman Institute, University of Illinois.
>> On Sun, Jun 30, 2019 at 10:06 AM WT Ren <> wrote:
>>> Dear All:
>>> I have been struggling to optimize the partial charges of one small
>>> molecule obtained from CHARMM-GUI by fftk and gaussian for several days. I
>>> followed the workflow in the tutorial, but unfortunately, the MM/QM
>>> distances and energies did not converge. Some water molecules moved far
>>> from the interaction site. I tried excluding those water molecules from the
>>> optimization, then other water molecules would move away. It seems there
>>> are some tricks in the charge optimization. Could anyone give me some
>>> advice or tips on this issue. I will be very grateful for any guidance. I
>>> can send the files if they would help.
>>> PS: The ligand molecule includes two phosphates and carries -4 charges.