Low coord low MW tmQM

[jaclark5]

New Submission Checklist

• Created a new folder in the submissions directory containing the dataset
• Added README.md describing the dataset see here for examples
• All files used to produce the dataset are included with a description
• [NA] Dataset follows the QCSubmit schema defined for Datasets, OptimizationDatasets and TorsionDriveDatasets
• Dataset filename matches pattern dataset*.json; may feature a compression extension, such as .bz2
• [NA] A PDF depicting the molecules is attached, in the case of torsiondrives this should include the highlighting of the central bond, this can be done automatically using qcsubmit.
• QCSubmit validation passed
• Made a new dataset entry in the mapping table in repository README.md
• Ready to submit!

[jaclark5]

@chrisiacovella before we merge this I'll need links to the source HDF5 for full posterity. I think there's enough here for @lilyminium to take a pass at reviewing, with the knowledge that links to the files will be provided.

[openff-dangerbot]

QCSubmit Validation Report

	submissions/2025-08-14-tmQM-xtb-Dataset-T=100K-Pd-Zn-Fe-Cu-low-mw-v0.0/scaffold.json.bz2
Dataset Name	tmQM xtb Dataset T=100K Pd Zn Fe Cu low mw v0.0
Dataset Type	singlepoint
Elements	F ,N ,Cl ,Cu ,Pd ,S ,P ,H ,Fe ,Zn ,Br ,C ,O
Valid Cmiles	❌
Connected Dihedrals	🔥
No Linear Torsions	🔥
No Molecular Complexes	❔
Valid Constraints	❔
Total Charge	🔥
Valid Coordinates	🔥
Complete Metatdata	❌

QC Specification Report

	submissions/2025-08-14-tmQM-xtb-Dataset-T=100K-Pd-Zn-Fe-Cu-low-mw-v0.0/scaffold.json.bz2/BP86/def2-TZVP
Specification Name	BP86/def2-TZVP
Method	bp86
Basis	def2-tzvp
Wavefunction Protocol
Implicit Solvent
Keywords	{"maxiter": 500, "scf_properties": ["dipole", "quadrupole", "wiberg_lowdin_indices", "mayer_indices", "lowdin_charges", "mulliken_charges"], "function_kwargs": {"properties": ["dipole_polarizabilities"]}}
Validated	🔥
Valid SCF Properties	🔥
Full Basis Coverage	🔥

QCSubmit version information(click to expand)

	version
openff.qcsubmit	0.56.0
openff.toolkit	0.17.0
basis_set_exchange	0.11
qcelemental	0.28.0
rdkit	2025.03.5

[openff-dangerbot]

QCSubmit Validation Report

	submissions/2025-08-14-tmQM-xtb-Dataset-T=100K-Pd-Zn-Fe-Cu-low-mw-v0.0/scaffold.json.bz2
Dataset Name	tmQM xtb Dataset T=100K Pd Zn Fe Cu low mw v0.0
Dataset Type	singlepoint
Elements	F ,N ,Cl ,Cu ,Pd ,S ,P ,H ,Fe ,Zn ,Br ,C ,O
Valid Cmiles	❌
Connected Dihedrals	🔥
No Linear Torsions	🔥
No Molecular Complexes	❔
Valid Constraints	❔
Total Charge	🔥
Valid Coordinates	🔥
Complete Metatdata	🔥

QC Specification Report

	submissions/2025-08-14-tmQM-xtb-Dataset-T=100K-Pd-Zn-Fe-Cu-low-mw-v0.0/scaffold.json.bz2/BP86/def2-TZVP
Specification Name	BP86/def2-TZVP
Method	bp86
Basis	def2-tzvp
Wavefunction Protocol
Implicit Solvent
Keywords	{"maxiter": 500, "scf_properties": ["dipole", "quadrupole", "wiberg_lowdin_indices", "mayer_indices", "lowdin_charges", "mulliken_charges"], "function_kwargs": {"properties": ["dipole_polarizabilities"]}}
Validated	🔥
Valid SCF Properties	🔥
Full Basis Coverage	🔥

QCSubmit version information(click to expand)

	version
openff.qcsubmit	0.56.0
openff.toolkit	0.17.0
basis_set_exchange	0.11
qcelemental	0.28.0
rdkit	2025.03.5

[lilyminium]

Mostly LGTM, thanks Jen! My only question is why there are 471097 conformers but 480339 entry names in the dataset? Are there duplicates?

Approving so as to not block if I've missed something obvious.

In the future could you please add a semicolon to suppress output when adding entries -- 400k integers adds a lot of scrolling to the notebook!

Edit: also, why is CI failing on cmiles?

[jaclark5]

Mostly LGTM, thanks Jen! My only question is why there are 471097 conformers but 480339 entry names in the dataset? Are there duplicates?

@lilyminium There are duplicate initial coordinates but at a different multiplicity. I supposed those will converge to a different geometry, so I could report it as 480,339 conformers.

In the future could you please add a semicolon to suppress output when adding entries -- 400k integers adds a lot of scrolling to the notebook!

Thanks for pointing this out! I didn't see that locally or in the files changed view on GitHub, I'll be sure to look at the raw file in the future.

Edit: also, why is CI failing on cmiles?

It's because TMOS is not always able to make a valid SMILES string, so those entries are without, causing an error.

[lilyminium]

I see -- thanks! Yeah I think 480339 is clearer and lets users know what dataset size to expect.