Assigning Legacy atom types and output topology

[amin-sagar]

Hello Everyone.

Sorry if this is a trivial question. I have seen the atom typing recovery example but I can't understand how to take just one molecule (a modified peptide) and write out the atom types as determined by the neural network.

Can someone please share an example minimal script to just take a single mol file or SMILES etc, assign gaff atom types and write out this information in some format.

What I am trying to do is to find a way to do hydration analysis on protein-ligand complexes using waterkit after parameterizing the ligand (small molecule or a peptide) using espaloma.

As described here, I can save the espaloma parameterized systems as amber files and use MDAnalysis to convert them to pdbqt files.

The issue is that the atomtypes saved in this manner are not understood by autodock tools.

So, what I would like to do is to assign legacy atom types to the peptides (which are covalently modified) using espaloma atom typing and retain the espaloma atom charges. This can be easily done by just replacing the last column of the pdbqt files with the predicted atom types.

I would be really grateful for any suggestions and help.

Best,
Amin.