Multiwfn forum

Abdirahman

Member

Registered: 2026-06-04

Posts: 1

Multiwfn MK RESP calculations

Hello, I need assistance, please.

So I am trying to replicate this study on MCPB.py, which metal parameter builder for metalloproteins. https://ambermd.org/tutorials/advanced/ … mcpbpy.php

The step I am currently having issues with is this step:
"Perform the Merz-Kollman RESP charge calculation for the large model". The MCPB.py natively supports Gaussian or Gamess. I don't have access to this QM software, so being utilising ORCA instead.

Hence, the reason me downloading Multiwfn to assist with MK RESP calculation.

To give some context:

The GAMESS input file consists of the following:

" $SYSTEM MEMDDI=400 MWORDS=200 $END
$CONTRL DFTTYP=B3LYP ICHARG=1 MULT=1 $END
$ELPOT IEPOT=1 WHERE=PDC $END
$PDC PTSEL=CONNOLLY CONSTR=NONE $END
$BASIS GBASIS=N31 NGAUSS=6 NDFUNC=1 $END
$DATA
Cluster/6-31G(d)
"

ESP charges are requested in GAMESS through the $ELPOT group. The keywords to perform ESP charge calculation are IEPOT=1 WHERE=PDC in this group. The first keyword turns on electrostatic potential calculation, and the second keyword requests the use of grids for the calculation.

The grid options are present in $PDC group. The two main keywords in this group are PTSEL and CONSTR. The first keyword gives the choice of the grid. The defaults are PTSEL=GEODESIC, which uses an algorithm by Mark Spackman, and CONSTR=CHARGE, which means the partial charges are fitted to the total charge of the system, too.
For the original method of Merz, Singh and Kollman, PTSEL=CONNOLLY is used.

Once the MK RESP calculations is complete, a log file is created which feeds into MCPB.pt step 3. which creates mol2 files, where these charges in these files are refitted by the MK RESP charge fitting algorithm.

For example, when atoms I am interested in Zn, we get charge from the MCPB.py tutorial on protein 10KL:
Zn = 0.433482, which binds to 3 histidine residues and the MNS ligand.

###

So I did a bit of experimentation to replicate these charges using Multifwn:

What I first did was run ORCA calculations on my model, based on these outputs, I generated a molden input file, which i then input into Multitwfn:

- following the examples given in manual Section 4.7.7.
I chose 7 , 18 , 1 for RESP calculations.  The .chg file is obtained.

The charge I get for the Zn  =  0.6275237761

I wonder where the discrepancy is, or doesn't matter.  I think this is important, as currently I am trying to run QM/MM; during the last stages of tleap, topology building for molecular dynamics, my charge on my system is not an integer, it float value. So that means it will be physically difficult to neutralise such a system with a non-integer charge value. I believe the main culprit is the way the charge is calculated for atoms.

Any suggestions, tips, please.

sobereva

Tian Lu (Multiwfn developer)

From: Athens, Greece

Registered: 2017-09-11

Posts: 2,243

Website

Re: Multiwfn MK RESP calculations

I am not a MCPB.py user, what I can answer is the way of using Multiwfn to calculate RESP charge.
I don't well understand your question. After loading a wavefunction file into Multiwfn, inputting 7 then 18 then 1 is the most straightforward way of calculating the RESP charge in common sense.

When calculating MK charges or RESP charges based on MK fitting points, different programs may use different density of fitting points and different atom radius for Zn (its radius was not directly defined in the original paper of MK method), the resulting charges may be notably different in some cases. The implementation of RESP charge calculation in Multiwfn is in a very reasonable way.