Error trying to run acpype with mopac

[gVallverdu]

Hi

I have a radical molecule and I would try to run acpype. As sqm does not handle open shell system I try to use mopac instead but I got errors.

• ACPYPE version is 2023.10.27
• AnteChamber 2023

First of all, I've installed mopac from conda. The mopac binary is mopac but acpype seems to look for mopac.sh. Do I need a specific mopac file?

Second, acpype doesn't set any input file for mopac (but it does for sqm). Maybe I have to provide this file? Basically the error comes from missing mopac input file.

Thank you in advance for any help.

Here is the output:

============================================================================
| ACPYPE: AnteChamber PYthon Parser interfacE v. 2023.10.27 (c) 2025 AWSdS |
============================================================================
DEBUG: CLI: -i input.pdb -b PVO -c bcc -a gaff2 -o gmx -n 0 -m 2 -q mopac
DEBUG: Python Version 3.12.3 | packaged by conda-forge | (main, Apr 15 2024, 18:38:13) [GCC 12.3.0]
DEBUG: Max execution time tolerance is 3h
DEBUG: setResNameCheckCoords done
DEBUG: Net charge drift '0.000000'
==> ... charge set to 0
==> ... converting pdb input file to mol2 input file
DEBUG: /softs/contrib/apps/anaconda/3/envs/AmberTools23/bin/obabel -ipdb input.pdb -omol2 -O PVO.mol2
==> * Babel OK *
==> Executing Antechamber...
DEBUG: /softs/contrib/apps/anaconda/3/envs/AmberTools23/bin/antechamber -dr no -i PVO.mol2 -fi mol2 -o PVO_bcc_gaff2.mol2 -fo mol2 -c bcc -nc 0 -m 2 -s 2 -df 0 -at gaff2 -pf n
DEBUG: No file left to be removed
ERROR: ++++++++++start_quote+++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
ERROR:
Welcome to antechamber 22.0: molecular input file processor.

Info: The atom type is set to gaff2; the options available to the -at flag are
 gaff, gaff2, amber, bcc, and sybyl.

Info: Finished reading file (PVO.mol2); atoms read (38), bonds read (45).
Info: Determining atomic numbers from atomic symbols which are case sensitive.
Running: /softs/contrib/apps/anaconda/3/envs/AmberTools23/bin/bondtype -j full -i ANTECHAMBER_BOND_TYPE.AC0 -o ANTECHAMBER_BOND_TYPE.AC -f ac
Warning: For atom (ID: 37, Name: V37) the best APS is not zero.
 Bonds involving this atom are frozen.

Running: /softs/contrib/apps/anaconda/3/envs/AmberTools23/bin/atomtype -i ANTECHAMBER_AC.AC0 -o ANTECHAMBER_AC.AC -p gaff2
Info: Total number of electrons: 191; net charge: 0
Info: The number of electrons is odd (191).
 Please check the total charge (-nc flag) and spin multiplicity (-m flag).

Running: /softs/contrib/apps/anaconda/3/envs/AmberTools23/bin/mopac.sh

 MOPAC is a semiempirical quantum chemistry program

 It uses a single argument, the MOPAC data-set
 The command to run MOPAC is 'mopac <data-set>.mop'

 For more information, see: HTTP://OpenMOPAC.net/Manual/running_MOPAC.html

 Press (return) to continue
/softs/contrib/apps/anaconda/3/envs/AmberTools23/bin/wrapped_progs/antechamber: Fatal Error!
Cannot open file (mopac.out) with mode (r).
No such file or directory

ERROR: ++++++++++end_quote+++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
ERROR: Antechamber failed
DEBUG: /softs/contrib/apps/anaconda/3/envs/AmberTools23/bin/parmchk2 -i PVO_bcc_gaff2.mol2 -f mol2 -o PVO_AC.frcmod -s 2

[alanwilter]

The mopac.sh lookup is a red herring — the real problem is more basic. ACPYPE ships a trimmed AmberTools, and neither mopac nor divcon is part of it:

$ ls acpype/amber_*/bin | grep -E 'mopac|divcon|sqm'
sqm

That is not something we stripped out: conda-forge's ambertools package does not contain them either. Only sqm is available. So -q mopac and -q divcon are advertised by the CLI but cannot work with a bundled install, which is a documentation bug on our side and I am sorry it cost you time.

What actually happens is that -q mopac makes ACPYPE pass -df 0 to antechamber; antechamber then tries to invoke its own mopac.sh wrapper, which is absent, and you get the missing-input-file error you saw. Installing conda's mopac binary does not satisfy it, because antechamber wants the AmberTools wrapper script rather than the executable.

For an open-shell radical, the options are:

1. Set the multiplicity and stay with sqm. You already pass -m 2; sqm will do a spin-unrestricted calculation. Worth checking whether it converges before reaching for mopac.
2. Compute charges externally and feed them in with -c user from a mol2 carrying your charges. This is the usual route for anything sqm cannot handle, and gives full control over the QM method.
3. Use a full AmberTools install that provides the mopac wrapper, with AMBERHOME pointed at it, rather than the bundled copy.

We will make the CLI help state that mopac and divcon need an external AmberTools. Leaving this open until that is documented.

[alanwilter]

Following up on the last point — this is now documented, in 482edcc.

acpype -h states it on the option itself:

--qprog  -q  <mopac|sqm|divcon>  am1-bcc engine; only sqm is bundled, mopac
                                 and divcon need an external AmberTools
                                 [default: sqm]

and the help footer now carries the full explanation, including the -c user route for an open-shell system sqm cannot converge. The README says the same in its options section.

Leaving this open in case you want to come back on how the radical itself worked out — whether sqm with -m 2 converged for you, or whether you went the external-charges route. Either data point is useful.