Voxel descriptors - problem with assigning atom types to SmallMol about moleculekit HOT 5 CLOSED

Hi Larry, I'll look into it the coming week. Thanks for the report :)

from moleculekit.

From the code side it looks fine. If you have VMD use this code to visualize the features in VMD, and also use SmallMolLib if you want.

from moleculekit.smallmol.smallmol import SmallMol
from moleculekit.smallmol.smallmollib import SmallMolLib
from moleculekit.tools.voxeldescriptors import getVoxelDescriptors, viewVoxelFeatures
import matplotlib.pyplot as plt
from mpl_toolkits.mplot3d import axes3d

# read molecule
suppl = SmallMolLib("GDP_ideal.sdf")
mol = suppl[0]

# generate voxels
features, centers, N = getVoxelDescriptors(mol,
                                     boxsize=(18, 16, 16),
                                     voxelsize=0.5,
                                     center=mol.getCenter())
viewVoxelFeatures(features, centers, N, voxelsize=0.5)
mol.view(style='Licorice')

As for the hydrophobicity and negative ionizable, @cuzzo87 will have to chime in as I'm no chemist. I assume it's an issue of rdkit though as you said

from moleculekit.

According to @cuzzo87 it's an rdkit issue since it can only return one descriptor/feature per atom. This means that an atom can be either hydrophobic or aromatic but not both. This is not something we can correct easily. But from an ML point of view it might not matter in the end as long as it's consistently putting atoms into one category.

from moleculekit.

Hi @stefdoerr @cuzzo87 thanks for looking into into this issue. I agree this should not be a problem for ML.

from moleculekit.

I will close this as we are not planning to modify rdkit. Thanks for the input!

from moleculekit.