Last post, I showed how to visualize each fingerprint contribution for predictive model results with RDKit.
The function also can visualize atomic properties such as a partial charge of atoms.
Dr. Thomas Evangelidis shared me an example of visualization of quantum chemistry based atom property. He showed an example of conformation effect to the QM calculation.
He tried to calculate PM-6 semi empirical partial charge and draw it with rdkit. He used MOPAC as QM engine. Following figure shows partial charge of same molecule with different conformation. Red arrow shows difference of each molecule.
And also by using partial change from 100 conformers data, he showed how to visualize std and mean of the partial charge.
From above image, conjugated aromatic system shows large difference between two conformers. It seems reasonable I think.
You can check whole his code on following URL.
Back to the latest rdkit blog post, Greg showed an example of visualization of partial charge with Extended Huckel method which is newly implemented on RDKit!
I had interest the approach so I tried to use psikit for partial charge calculation.
Following code is almost same as original blog post but little difference, just using psikit.
Current version of psikit can calculate mulliken charge, esp, resp charge and lowdin charge very conveniently. But it took longer time for calculation compared to rdEHTools method.
psikit seems more sensitive to aromatic atoms. I think it is needed to select method in case by case. But rdkit’s rdEHTools works very fast so it is useful tool of light weight QM calc.
Finally thanks Dr. Thomas again for sharing nice code example!