Dear David,
Thank you for your answer.
"Can you clarify what you mean by "the first frame of the QM/MM simulation"? Did you do a one- or two-step minimization, and look at the output coordinates?"
Yes, this problem persists after several tens of minimization steps (>50). What I mean by the "first frame of the QM/MM minimization" is that the water molecule is already distorted in the first QM geometry generated by the Amber/ORCA interface (the generated file inpfile.xyz). The water molecule structure is correct in the pdb (actually .rst) I use to start the QM/MM minimization. That's why I am paying attention to the QM/MM module. I have also checked the following:
1) The QM structure of my ligand is correct. Thus, this problem only affects to the water molecule not covalently linked to my ligand.
2) If a second water molecule is added to the QM partition, having thus the ligand + 2 WAT, both water molecules are distorted by the QM interface.
3) If take the ligand out of the QM partition and I keep only one water molecule, the coordinates of the water molecule are correct in the QM input.
4) I have tried other QM programs such as Gaussian 16. Same problem. I have also tried other versions of Amber, in particular, Amber 18. Same result. Therefore, it can be concluded that this issue is not related to a specific Amber/QM interface or Sander version, but rather to the &qmmm module of Amber.
5) This issue is somewhat related to the keyword qmshake. My previous comments (1-4) refer to qmshake=1. If qmshake=0, the coordinates of the water molecules are not distorted at the beginning of the qm/mm runs and look fine, but unphysical distortions take place in the QM/MM dynamics after several hundreds of steps (or even in the previous minimizations). In particular, what I observe is sudden O-H bond enlargements (> 6 Angstrom) of the water molecule. This motivated the use of qmshake=1, which in turn gave me the problems I'm explaining here.
"QM/MM folks: is there some way to save or otherwise access the Orca input file?"
If I understood well your question, you can find the Orca input file in the directory where you execute the QM runs (usually the same where you execute Sander). The name is "orc_job.inp" for the current step, and old.orc_job.inp for the previous step. Same for outputs (*.dat instead of *.inp).
Best regards,
Antonio
On Tuesday, January 18, 2022 13:42 CET, David A Case <david.case.rutgers.edu> wrote:
> On Mon, Jan 17, 2022, Antonio Frances Monerris wrote:
> >
> >I am performing QM/MM simulations with Amber 20 and Orca 5.0. I would like
> >to report a problem when including water molecules in the QM partition
> >using Sander (Amber 20).
> >
> >The water molecules inside the QM partition appear with wrong H-O-H angles
> >(~70 ยบ) and short O-H distances (~0.7-0.8 A) already in the first frame of
> >the QM/MM simulation. The water molecules minimized with MM had correct
> >parameters prior to the QM/MM job. This strange water molecule distortions
> >are not solved during the QM/MM minimization.
>
> Can you clarify what you mean by "the first frame of the QM/MM simulation"?
> Did you do a one- or two-step minimization, and look at the output
> coordinates? Try to find the simplest calculation that you can that> illustrates the problem.
>
> QM/MM folks: is there some way to save or otherwise access the Orca input
> file?
>
> ....dac
>
>
> _______________________________________________> AMBER mailing list
> AMBER.ambermd.org
> http://lists.ambermd.org/mailman/listinfo/amber
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Tue Jan 18 2022 - 08:30:02 PST
