Merge pull request #51 from RPirie96/fix_lipinski

Using the original definitions for HBD/HBA calc for Lipinski

CHANGELOG

@@ -8,6 +8,8 @@ Authors: Mateusz K. Bieniek, Ben Cree, Rachael Pirie, Joshua T. Horton, Natalie
  - The user now has to import the libraries (RGroupGrid) and instantiate first.
  - Libraries (linkers, rgroups) are now single .sdf files to avoid problems on clusters with many small files
  - A growing vector can now be any molecule-separating atom in the molecule.
+ - Lipinski now uses the initial definitions of HBD: "sum of Ns and Os" and HBA: "sum of OHs and NHs",
+    (thanks to @RPirie96)
  - When growing a molecule C multiple time in the same session (e.g. +linker +R), the .template attribute will
     always be the original C
 

fegrow/toxicity.py

@@ -1,5 +1,5 @@
 import pandas as pd
-from rdkit.Chem import Descriptors
+from rdkit.Chem import Descriptors, Lipinski
 from rdkit.Chem.FilterCatalog import FilterCatalog, FilterCatalogParams
 
 from .sascorer import calculateScore
@@ -16,8 +16,8 @@ def rule_of_five(mol):
 
     # Ro5 descriptors
     MW = Descriptors.MolWt(mol)
-    HBA = Descriptors.NumHAcceptors(mol)
-    HBD = Descriptors.NumHDonors(mol)
+    HBA = Lipinski.NOCount(mol)
+    HBD = Lipinski.NHOHCount(mol)
     LogP = Descriptors.MolLogP(mol)
 
     # Ro5 conditions