Closed BJWiley233 closed 1 year ago
Works on my end, see PDBQT attached. What OS, and versions of Python, Meeko, and RDKit are you using? meeko-issue-41.pdbqt.txt
Let me get that info for you.
ubuntu 20.04 Python 3.7.9 Meeko 0.4.0 rdkit 2022.09.1
Worked on my Mac as well with:
Python 3.9.13 Meeko 0.3.3 rdkit 2022.03.5
The nans and infs are coming from rdkit 2022.09, but not from 2022.03, although I don't think it has to do with the rdPartialCharges
module, but with the way the phosphate is perceived by rdkit (the phosphate in your molecule has a weird geometry). Here are the smiles produced by rdkit's Chem.MolToSmiles(mol)
for the phosphorous atom:
[P@TB4]
2022.03.1 P
Looks like the newer rdkit 2022.09.1 is labeling that P with some chiral flags that are inferred from the (incorrect) phosphate geometry. Manually editing the coordinates for that phosphate to a tetrahedral geometry resolves the issue and works fine with rdkit 2022.09.1 (see attaached file, which differs from the one you posted only at the coordinates for one O and one H). tetrahedral-phosphate.sdf.txt
For this sdf file generated from smiles I am getting
nan
for most of the charges but not all of them. Obviously autogrid and thus autodock won't work with this.example from pdbqt
sdf