Impact of 'useBondTypes' on GetMorganFingerprint()

Cheminformatics Tutorials - Herong's Tutorial Examples

∟Impact of 'useBondTypes' on GetMorganFingerprint()

This section provides a tutorial example on impact of the 'useBondTypes' option on fingerprint generation with rdkit.Chem.rdMolDescriptors.GetMorganFingerprint() function.

The 'useBondTypes' option in the rdkit.Chem.rdMolDescriptors.GetMorganFingerprint() function call allows you to control whether or not bond types (bond orders) are considered, when updating identifiers with neighboring atom nodes. The default value is useBondTypes=True.

1. For example, molecule "C=C" has a double bond. The identifier 3695448525 in the fingerprint represents atom C with a neighboring C connected by a double bond with the useBondTypes=True option.

from rdkit.Chem import AllChem
from rdkit.DataStructs.cDataStructs import UIntSparseIntVect

radius = 1
bitInfo = {}
mol = AllChem.MolFromSmiles('C=C')
fp = AllChem.GetMorganFingerprint(mol, radius, useBondTypes=True, 
  bitInfo=bitInfo)
display(UIntSparseIntVect.GetNonzeroElements(fp))
display(UIntSparseIntVect.GetTotalVal(fp))
print(bitInfo)

# output: 
{2246997334: 2, 3695448525: 1}
3
{2246997334: ((0, 0), (1, 0)), 3695448525: ((0, 1),)}

Note that I am expecting identifier 3695448525 to appear 2 times in the fingerprint, since the 2 C atoms have identical local structures.

2. If we use useBondTypes=False, The identifier representing atom C with a neighboring C connected by a double bond changes to 3695449228.

radius = 1
bitInfo = {}
mol = AllChem.MolFromSmiles('C=C')
fp = AllChem.GetMorganFingerprint(mol, radius, useBondTypes=False, 
  bitInfo=bitInfo)
display(UIntSparseIntVect.GetNonzeroElements(fp))
display(UIntSparseIntVect.GetTotalVal(fp))
print(bitInfo)

# output: 
{2246997334: 2, 3695449228: 1}
3
{2246997334: ((0, 0), (1, 0)), 3695449228: ((0, 1),)}

Conclusion: The 'useBondTypes' option is only used during the updating identifier phase. It has no impact on the initial identifier generation phase. We should keep the default of useBondTypes=True to encode bond types into updated identifiers.