Systematic approach to parameterise covalently bound ligand

[JSLJ23]

Just wondering, is there a best practice approach and/or a systematic way to go about running OpenMM simulations for covalently bound ligands to protein residues?
I understand that the restriction of classical force-fields are unable to parameterise the covalently modified residue, hence I was wondering if there was a way to:

1. Use espaloma to build a molecular graph of the covalently modified residue, and perhaps a neighbourhood of 2-3 other residues in the N and C-terminal direction to get better context of the chemical environment
2. Assign parameters for the whole molecular graph.
3. Extract the required parameters only for the modified residue and add that to the force field parameterisation by Amber14 for the system.