'NoneType' object has no attribute 'coordinates'

[JonasLi-19]

I have read the tutorial but there are something wrong with my code:

from spyrmsd import io, rmsd, molecule

...
            ref_pose = os.path.join(crystal_dir, pdb_id, f'{pdb_id}_ligand.sdf')
            ref = io.loadmol(ref_pose)
            mol = io.loadmol(new_filepath)

            ref = ref.strip()
            mol = mol.strip()

            coords_ref = ref.coordinates
            anum_ref = ref.atomicnums
            adj_ref = ref.adjacency_matrix

            coords = mol.coordinates
            anum = mol.atomicnums
            adj = mol.adjacency_matrix

            symm_rmsd_value = rmsd.symmrmsd(
                coords_ref,
                coords,
                anum_ref,
                anum,
                adj_ref,
                adj,
            )

Then it raises ERROR:

Traceback (most recent call last):
  File "analysis_docked_poses_score_to_csv.py", line 70, in <module>
    coords_ref = ref.coordinates
AttributeError: 'NoneType' object has no attribute 'coordinates'

[RMeli]

This error indicates that the ref object is None. This happens if the molecule is not properly loaded. What are you using as a backend? rdkit does return None if a molecule can't be parsed/sanitized. Do you get any warnings/error from RDKit/Open Babel when loading the molecule?

[JonasLi-19]

I thought if I use io.load, then I don't need openbabel/RDKit?

The spyrmsd.io module provides functions to easily load molecules from a file and to transform them into a spyrmsd.molecule.Molecule object:

ref = io.loadmol("molecules/1a4k_ligand.sdf")

This is the code I tried using io.load(), without openbabel/rdkit:

from spyrmsd import io, rmsd, molecule

...
            ref_pose = os.path.join(crystal_dir, pdb_id, f'{pdb_id}_ligand.sdf')
            ref = io.loadmol(ref_pose)
            mol = io.loadmol(new_filepath)

            ref = ref.strip()
            mol = mol.strip()

            coords_ref = ref.coordinates
            anum_ref = ref.atomicnums
            adj_ref = ref.adjacency_matrix

            coords = mol.coordinates
            anum = mol.atomicnums
            adj = mol.adjacency_matrix

[JonasLi-19]

And this link is the code I tried with RDKit, but still error: #97 (reply in thread)
There is indeed some warning with the molecule:

[20:23:05] Warning: molecule is tagged as 2D, but at least one Z coordinate is not zero. Marking the mol as 3D.
[20:23:05] Warning: molecule is tagged as 2D, but at least one Z coordinate is not zero. Marking the mol as 3D.
Traceback (most recent call last):
  File "analysis_docked_poses_score_to_csv.py", line 82, in <module>
    coords_ref = ref_molecule.coordinates
AttributeError: 'NoneType' object has no attribute 'coordinates'

[RMeli]

spyrmsd.io still needs either RDKit or OpenBabel to be installed. My personal preference is for RDKit, which has stricter checks. spyrmsd does not implement parsing molecular files, that's what RDKit and Open Babel do well already.

Once you have RDKit installed, you can use io.load to load the molecules, without having to use RDKit directly yourself (as you did in #97 (reply in thread)). So, once RDKit is installed the code should work. The error you see is likely caused by issues with your file, which can't be properly parsed by RDKit.

I haven't seen this issue before, but I suspect that coordinates are not properly included given that the molecule is tagged as 2D?

[RMeli]

BTW, if you have spyrmsd molecules already, you can use the rmsdwrapper function instead of passing the information directly to symmrmsd. I should document this better for sure.

from spyrmsd import io, rmsd, molecule

...
            ref_pose = os.path.join(crystal_dir, pdb_id, f'{pdb_id}_ligand.sdf')
            ref = io.loadmol(ref_pose)
            mol = io.loadmol(new_filepath)

            reds = rmsdwrapper(ref, mol) # By default, strips hydrogens and use symmetry-correction

spyrmsd/spyrmsd/rmsd.py

Lines 300 to 308 in 8d3c540

	def rmsdwrapper(
	molref: molecule.Molecule,
	mols: Union[molecule.Molecule, List[molecule.Molecule]],
	symmetry: bool = True,
	center: bool = False,
	minimize: bool = False,
	strip: bool = True,
	cache: bool = True,
	) -> Any:

[RMeli]

@Dadiao-shuai, I got an e-mail notification with your question, but I can't see it anymore on GitHub. Did you remove it? (Please don't. It's a bit confusing for me, and every question is valuable for the community.)

---

The RDKit error

ERROR: Could not sanitise molecule ending on line 102
ERROR: Explicit valence for atom # 6 C, 6, is greater than permitted

means that RDKit can't sanitise the molecule because of problems with your input file (specifically, there is a carbon with 6 bonds). This leads to the loaded molecule being None, hence the NoneType error.

I'll try to provide a better error message when molecules can't be loaded by RDKit or OpenBabel instead of having Nones around.