Suggestion for a better way to protonate ligands.

[dargen3]

Hello,

I think hydride has great potential for its versatility and speed. I have tried many tools for adding hydrogens from structures over the last months, but hydride is the best for me, together with the pdb2pqr tool. The weakness of hydride (and also other protonation tools) is the protonation of ligands, which it always protonates to a neutral state. Hydride has implemented an option to recalculate formal charges by pKa for standard residues. I propose to implement a similar approach for other functional groups, such as openbabel. The problem with openbabel is that it doesn't work well with structures without defined bonds (PDB, mmCIF), which is hydride's strength. The approach is already implemented for residues, so extending it to other functional groups might not be that much work and it would significantly increase the quality of ligand protonation.

If it would be too much work or you don't have time to do it then please give me a hint and I can try to implement this feature myself sometime in the future.

[padix-key]

This sounds like a great idea! Indeed I do not have much spare time, so if you prepare a PR re-implementing the charge assignments from OpenBabel (maybe a function called estimate_charges(atoms, ph)?) this would be really nice.

[dargen3]

Okay, but it's more likely to be in the future. Okay, but it's more likely to be in the future. As I think about it, it could be a nice topic for a bachelor's thesis.

[dargen3]

Hello,

I'm thinking about implenetation and I would like to ask if the function can use other libraries? I'm thinking now specifically of RDKit, because it is suitable for working with small molecules and can work great with the SMILES format, which is essential for the implementation.

[padix-key]

I am a bit hesitant to add another dependency, especially as RDKit is rather heavy weight. You probably need the SMILES strings to match the groups from https://github.com/openbabel/openbabel/blob/master/data/phmodel.txt, right? Is there a way to represent them differently, i.e. to convert the SMILES strings with RDKit into some format Biotite can use directly and then ship this format instead?

If there is no reasonable way around it, I think introducing RDKit is okay, but rather as optional dependency in the pyproject.toml, so the rest of Hydride still works in environments where RDKit is not installed.