protein-param-fit

OpenFF force field parameter fits for proteins

Manifest

• 1-curate-training-datasets.py. Pulls down QC training datasets from QCArchive.
• 2-check-elf10-charges.py. Filters molecules in the training dataset for which ELF10 charge assignment fails.
• conda-envs. Directory containing minimal and solved conda environments.
• delocalized-charge-smirks. Directory containing a force field with parameter SMIRKS for delocalized charges but no protein library charges.
• library-charges. Directory containing the derivation of library charges for protein residues.
• null-model. Directory containing the parameter fit for the null model, which uses Sage and protein QC training data but has no protein-specific SMIRNOFF types.
• sage-behara.offxml. Initial force field taken from Pavan Behara here
• test-delocalized-charge-assignments.py. Identifies molecules in a QC dataset that are assigned different parameters to chemically equivalent atoms with delocalized charges.
• training-datasets. Directory containing QC training datasets as JSON files.