As chemoinformatitians know ;) recently new version of rdkit is released. I appreciate the great work for developpers!
One of the interesting point for me is that, FreeWilson(FW) analysis was added to Contrib. FW analysis is a traditional approach of chemoinformatics but I think it’s really MedChem friendly approach. You can get lots of information about FW when googling “free wilson analysis” as a query.
So I tried to use FW analysis in RDKit/Conrib. At first, I updated my rdkit with mamba command.
$ mamba update -c conda-forge rdkit
After updating RDKit, I tested FW. **Most of following code is borrowed from original code.
The Redge regression is used for prediction method. If you don’t have scaffold definition, you can use rdFMCS to find the scaffold. How to use rdFMCS in FW analysis is well described in https://github.com/rdkit/rdkit/tree/master/Contrib/FreeWilson.
Finding MCS for large molecular sets will take long time. It requires computational cost.
I’m testing fastpages powered blog generation. The link is below.
https://iwatobipen.github.io/is_life_worth_living-/jupyter/2021/10/23/rdkitcontribfreewilson.html
After running the FWDecompose, decomp object has r-group and scaffold information. And it has some filters because generate all R-Group combinations will generate huge amount of molecules.
It’s really interesting work I think.
And also if you would like to run FW analysis from CLI, I recommend to check Pat’s cool repo. https://github.com/PatWalters/Free-Wilson
Is life worth living? 