Update aa_residues_substructures_explicit_bond_orders_with_caps_expli…

[pbuslaev]

This patch should solve #1727

[codecov]

Codecov Report

Merging #1728 (7c37650) into main (88ce0b3) will decrease coverage by 11.74%.
The diff coverage is n/a.

Additional details and impacted files

[j-wags]

Thanks for starting on the fix for this!

Patching that file will only solve the problem temporarily, since we periodically regenerate it from the chemical components dictionary using this script. Also, I'm seeing CYS residues in our current dictionary with a negative sulfur. So either there's something wrong with a existing substructure in this file that's causing it not to match the deprotonated CYS in the input, or the mainchain substructure is fully missing.

It'd be great if we could solve the problem by substituting an existing substructure in the _patch_known_problems function, but barring that, just adding a new working substructure in the _add_common_substructures function would be fine.

[pbuslaev]

I think that the changes I added should lead to generation of the correct json file. I have also updated the json file and fixed a typo in rdkit_wrapper docstring

[j-wags]

Wow - Thanks for diving into our crazy substructure code! The resulting patterns look good visually but I tried loading the PDB file from your original post and it failed with:

E           openff.toolkit.utils.exceptions.UnassignedChemistryInPDBError: Some bonds or atoms in the input could not be identified.
E           
E           Hint: The following residues were assigned names that do not match the residue name in the input, or could not be assigned residue names at all. This may indicate that atoms are missing from the input or some other error. The OpenFF Toolkit requires all atoms, including hydrogens, to be explicit in the input to avoid ambiguities in protonation state or bond order:
E               Input residue  :CYS#0002 contains atoms matching substructures {'No match', 'PEPTIDE_BOND'}
E           
E           Error: The following 5 atoms exist in the input but could not be assigned chemical information from the substructure library:
E               Atom     9 (HA) in residue  :CYS#0002
E               Atom    10 (CB) in residue  :CYS#0002
E               Atom    11 (HB2) in residue  :CYS#0002
E               Atom    12 (HB3) in residue  :CYS#0002
E               Atom    13 (SG) in residue  :CYS#0002

openff/toolkit/utils/rdkit_wrapper.py:818: UnassignedChemistryInPDBError

I tried shaking a few permutations of things like residue names, but to no avail. Are you able to load the original PDB successfully?

In either case, I have an example file and test we can hammer on, but I don't have write access to your fork to send it over. Would you be OK if I merged this into a branch so OpenEye CI stops complaining, and then gave you write access?

[pbuslaev]

Wow - Thanks for diving into our crazy substructure code! The resulting patterns look good visually but I tried loading the PDB file from your original post and it failed with:

E           openff.toolkit.utils.exceptions.UnassignedChemistryInPDBError: Some bonds or atoms in the input could not be identified.
E           
E           Hint: The following residues were assigned names that do not match the residue name in the input, or could not be assigned residue names at all. This may indicate that atoms are missing from the input or some other error. The OpenFF Toolkit requires all atoms, including hydrogens, to be explicit in the input to avoid ambiguities in protonation state or bond order:
E               Input residue  :CYS#0002 contains atoms matching substructures {'No match', 'PEPTIDE_BOND'}
E           
E           Error: The following 5 atoms exist in the input but could not be assigned chemical information from the substructure library:
E               Atom     9 (HA) in residue  :CYS#0002
E               Atom    10 (CB) in residue  :CYS#0002
E               Atom    11 (HB2) in residue  :CYS#0002
E               Atom    12 (HB3) in residue  :CYS#0002
E               Atom    13 (SG) in residue  :CYS#0002

openff/toolkit/utils/rdkit_wrapper.py:818: UnassignedChemistryInPDBError

I tried shaking a few permutations of things like residue names, but to no avail. Are you able to load the original PDB successfully?

This is strange. With updated json substructure file everything works perfectly for me. I do see the exact same SMARTS pattern as I added manually in the generated substructure file, so I am a bit confused why you are getting the error.

In either case, I have an example file and test we can hammer on, but I don't have write access to your fork to send it over. Would you be OK if I merged this into a branch so OpenEye CI stops complaining, and then gave you write access?

Yes, this is fine with me.