Conformer Screening

[hansen7]

Thanks for sharing!

I can't find the codes for the conformer screening, which is the descried in the arXiv:

We generate mutliple conformers for each molecule with RDKit [Landrum et al., 2006] and prune similar conformers with an RMSD cutoff R = 0.5Å.

neither from here

MoleculeX/Molecule3D/preprocess/PubChemQCDataset.py

Lines 195 to 197 in 29951d7

	smiles = Chem.MolToSmiles(mol)
	coords = mol.GetConformer().GetPositions()
	z = [atom.GetAtomicNum() for atom in mol.GetAtoms()]

nor here:

MoleculeX/molx/dataset/molecule3d.py

Lines 135 to 137 in 29951d7

	smiles = Chem.MolToSmiles(mol)
	coords = mol.GetConformer().GetPositions()
	z = [atom.GetAtomicNum() for atom in mol.GetAtoms()]

would you mind elaborating a bit more?

[floatlazer]

Hi,

Thanks for your interest. The paper you referred corresponds to our kddcup solution: https://github.com/divelab/MoleculeX/tree/molx/BasicProp/kddcup2021

The code for conformer generation is here:

MoleculeX/BasicProp/kddcup2021/conformer/gen_confs.py

Line 30 in 29951d7

cids = AllChem.EmbedMultipleConfs(mol, numConfs=40, pruneRmsThresh=0.5)