Re: QM/MM error in the charges

From: Pawel Kedzierski (
Date: Fri Feb 28 2020 - 08:06:44 CST

W dniu 28.02.2020 o 13:44, Aashish Bhatt pisze:
> Thank you sir for your response
> Can you please elaborate sir.
> I need to manually edit the charges or i am getting something wrong.

Manually editing charges is bad science, they describe interactions in
your system. You probably need to educate yourself more and then confirm
with your scientific tutor/supervisor whether what you plan to do is
sensible or not.

You were advised to change which atom compose your QM part in such a
way, that the partial charges of the entire QM region sum up to integer
value. How to do it, depends on the parameters. For biological residues
(amino acids, nucleotides) the charges should sum up to integer values
for entire residue in any force field, so the QM/MM division on residue
boundary should be safe. For arbitral molecules the charges will sum up
to an integer value (the net formal charge) for the whole molecule.

If one is calculating the parameters himself, the way is to add
constraints to charge fitting to get the specified sum of charges within
a fragment of the molecule. Such fragments define "charge groups" which
dr Vermaas refers to; in CHARMM FF, the biological residues are already
divided into relatively small charge groups which allows for better
granularity of division in QM/MM context.

It is possible to define charge groups using Amber Tools to parametrize
arbitral molecules. Look up the "respgen" chapter in Amber Tools Manual
for a start. Bear in mind, however, that if you impose nonsense
constraints you will get nonsense charges so in general, you should set
up a benchmark to test your parameters if they work well (that is,
reproduce the properties of interest).

With regards,


> Best Regards
> Aashish
> On Thu, Feb 27, 2020 at 8:31 PM Josh Vermaas <
> <>> wrote:
> Cut somewhere else between QM and MM regions. The errors are
> pretty clear, in that it demands integer charges so that PME
> works. Amber topologies do not have the "group" concept like
> CHARMM topologies do, so the atomic charges within a molecule
> likely don't add to an integer.
> -Josh
> On Thu, Feb 27, 2020 at 5:51 AM Aashish Bhatt
> < <>>
> wrote:
> Dear Sir/Ma'am
> I am trying to check bond breaking in my system with the help
> of QM/MM.
> I am getting the following error and i am using Amber topology.
> can you please look into it.
> *Info: 1) Group ID: 1 ; Group size: 9 atoms ; Total PSF
> charge: -0.412152
> FATAL ERROR: QM atoms do not add up to a whole charge, which
> is needed for PME.
> ------------- Processor 0 Exiting: Called CmiAbort ------------
> Reason: FATAL ERROR: QM atoms do not add up to a whole charge,
> which is needed for PME.*
> *
> *
> Best Regards
> Aashish

This archive was generated by hypermail 2.1.6 : Thu Dec 31 2020 - 23:17:12 CST