RDKit 2018.03.01 release! #rdkit

Dear RDKitter,
It’s good news that new version of rdkit is released!
You can find details in original repository.
https://github.com/rdkit
There are many improvement and bug fixes in the release. I appreciate developers work!
Recent version of RDKit has lots of 3D descriptors. PMI/NPR.
https://pubs.acs.org/doi/abs/10.1021/jp004042l?journalCode=jpcafh
And new version of rdkit has new function “ComputePrincipalAxesAndMoments” that computes principal axes and moments of inertia for a conformer.
I tried to use the function. ;-)
At first I installed new version of rdkit via conda command.

iwatobipen$ conda install -c rdkit rdkit=018.03.1.0

Then wrote code.

from pprint import pprint
from rdkit import Chem
from rdkit import rdBase
from rdkit.Chem import Draw
from rdkit.Chem.Draw import IPythonConsole
from rdkit.Chem import AllChem
print(rdBase.rdkitVersion)
# 2018.03.1
mol=Chem.MolFromSmiles('O=C2C(Cc1ccc(cc1)C(C(=O)O)C)CCC2')
mol = Chem.AddHs(mol)
#generate 3d conf
AllChem.EmbedMolecule(mol, useExpTorsionAnglePrefs=True, useBasicKnowledge=True)
#calc PMI and NPR with default function
pmi1=AllChem.CalcPMI1(mol)
pmi2=AllChem.CalcPMI2(mol)
pmi3=AllChem.CalcPMI3(mol)
npr1 = AllChem.CalcNPR1(mol)
npr2 = AllChem.CalcNPR2(mol)
pprint([pmi1, pmi2, pmi3,"##",npr1, npr2])
[Out]
[484.9333356244745,
 2954.3752956989097,
 3158.6098121028335,
 '##',
 0.15352745811348945,
 0.9353403780291699]

# Next use new function
from rdkit.Chem import rdMolTransforms as rdmt
rdmt.ComputePrincipalAxesAndMoments(mol.GetConformer())
[Out]
(array([[ 0.98500524,  0.17113466,  0.02185434],
        [-0.1185898 ,  0.57961538,  0.8062149 ],
        [-0.1253042 ,  0.7967176 , -0.59121901]]),
 array([ 30.83932592, 203.65737152, 218.01591889]))

There are differences in both function. Why ??? I checked source code but I can not understand the reason.
Most difference is scale, but not equal.

print([484.9333356244745/30.83932592, 2954.3752956989097/203.65737152, 3158.6098121028335/218.01591889])
[Out]
[15.724511517613434, 14.506596415582127, 14.487977887965654]

RDKit 3D conformer generator shows good performance as same as commercial packages. It indicates 3D descriptors of RDKit is useful in Computer Aided Drug Discovery I think.

At last, I attached today’s snippet as PDF files.
test1