[gmx-users] Link atoms issue: QMMM simulation

Zachary Smithline zachary.smithline at yale.edu
Fri Dec 1 06:51:02 CET 2017

Deal All,

Does anybody have experience doing QMMM simulations. I am interfacing
Gromacs 4.5.7 with Gaussian 09-C.01 via Gerrit Groenhof's gau script. I
have 2 questions:

1. I was performing a toy run for QMMM ONIOM using lysozyme. I was treating
the side chain of one of the lys residues as the QM layer. I was able to
successfully generate the .tpr via grompp, and start the md run. The
problem is that the Gaussian input files (input.com as produced by gau)
seem to be missing the link atom that connects the lys residue to the rest
of the protein. I have followed the procedure as listed here exactly for
setting up my simulation:

Do you know what could be wrong? Is it an error in the gau script?

2. Next, I went ahead and tried to perform a similar run to the one
described above with my real system, which is large and has 144 QM atoms
and 7 link atoms. This time, I did not even get past the grompp stage. The
error I get from grompp is:


Program grompp, VERSION 4.5.7

Source code file: vsite_parm.c, line: 797

Fatal error:

Automatic parameter generation not supported for Virtual site 2 atom 2458

For more information and tips for troubleshooting, please check the GROMACS

website at http://www.gromacs.org/Documentation/Errors


Does anybody know what could be causing this fatal error? Atom 2458 is the
first link atom, as listed in my .gro coordinate file. I looked online,
and, regarding a similar error (except pertaining to "Virtual site 3"), it
was suggested that constraints be set to all-bonds in the .mdp. My
constraints is definitely set to all-bonds.

Thank you in advance!

Zach Smithline

More information about the gromacs.org_gmx-users mailing list