Experiment

Experiment
Select single molecule and run FB optimization fitting only TIG0 (general torsion parameter)	The selected target is: td_OpenFF_Group1_Torsions_717_C7H9N3O -General torsion ID with wildcards in every bond and atom
Select single molecule and run FB optimization fitting only TIG-fit0 Smirks: [:1]~[#6X3:2]~!@[#6X3:3]~[:4]	Open Issues with parameter matching
Iterate through single molecule and run FB optimization fitting only TIG-fit0 Smirks for TIG-fit0: [:1]~[#6X3:2]~!@[#6X3:3]~[:4]	Fixed issue with parameter matching Workflow: Code is here: Workflow: Check if torsion drive matches the indices of the TIG-fit0 Run force balance fit for each torsion Plot the QM, MM iteration 0, MM iteration 1 of the torsions Open
	TIG-fit0, uses starting values from t44 parameter

Torsion Barrier versus k value:

A comparison of the torsion barrier and the kvalue. The torsion barrier is determined by an algorithm which checks the slope of a torsion drive record and pulls the maximum. The k value is determined by a single molecule force balance run that fits a general torsion which is applied 4x fold around a single torsion.

Substituted Phenyl Dataset:

Open

OpenFF 1.2.0 Training Datasets and Substituted Phenyl Dataset:

Perform steric filtering on molecules:

Determine the torsion ID using the entry_label from the metadata in the targets folder.

import qcportal as ptl
client = ptl.FractalClient()
ds = client.get_collection("TorsionDriveDataset", "OpenFF Substituted Phenyl Set 1")
ds.status(status="COMPLETE")
entry = ds.get_entry(entry_label)
entry.object_map['default']

Additional targets in fitting experiment:

Original fitting experiment used only the gen 2 torsions and the substituted phenyl dataset. These plots include also the primary benchmark dataset, group1 torsions, and the Rowley dataset.

Fitting experiment for non-ring torsions:

Steric filtered:

Non-steric filtered:

Fitting experiment for general C~C torsion:

Steric filtered:

Non-steric filtered: