GetRDKitFPGenerator() Method in RDKit

Cheminformatics Tutorials - Herong's Tutorial Examples

∟Daylight Fingerprint Generator in RDKit

∟GetRDKitFPGenerator() Method in RDKit

This section provides a quick introduction on the rdkit.Chem.rdFingerprintGenerator.GetRDKitFPGenerator() Method in the RDKit library.

GetRDKitFPGenerator() method is located in the rdkit.Chem.rdFingerprintGenerator module of the RDKit library. It creates a Daylight fingerprint generator that uses the Daylight method to generate integer identifiers from subgraphs in the molecular.

Here is the definition of the method:

rdkit.Chem.rdFingerprintGenerator.GetRDKitFPGenerator([
  (int)minPath=1 [, 
  (int)maxPath=7 [, 
  (bool)useHs=True [, 
  (bool)branchedPaths=True [, 
  (bool)useBondOrder=True [, 
  (bool)countSimulation=False [, 
  (AtomPairsParameters)countBounds=None [, 
  (int)fpSize=2048 [, 
  (int)numBitsPerFeature=2 [, 
  (AtomPairsParameters)atomInvariantsGenerator=None]]]]]]]]]]) 
-> FingerprintGenerator

Descriptions of method arguments are:

minPath (optional): the minimum path length (in bonds) to be included.
maxPath (optional): the maximum path length (in bonds) to be included.
useHs (optional): toggles inclusion of Hs in paths (if the molecule has explicit Hs).
branchedPaths (optional): toggles generation of branched subgraphs, not just linear paths.
useBondOrder (optional): toggles inclusion of bond orders in the path hashes.
useCountSimulation (optional): if set, use count simulation while generating the fingerprint.
countBounds (optional): boundaries for count simulation, corresponding bit will be set if the count is higher than the number provided for that spot.
fpSize (optional): size of the generated fingerprint, does not affect the sparse versions.
numBitsPerFeature (optional): the number of bits set per path/subgraph found.
atomInvariantsGenerator (optional): atom invariants to be used during fingerprint generation.

Once a FingerprintGenerator object is created, you can call its instance methods to generate different types of Morgan fingerprints of a given molecule.

gen.GetFingerprint(mol): returns a bit vector of size fpSize
gen.GetCountFingerprint(mol): returns a count vector of size fpSize
gen.GetSparseFingerprint(mol): returns a sparse bit vector
gen.GetSparseCountFingerprint(mol): returns a sparse count vector

The above methods are actually providing same functionalities as Morgan fingerprint generation methods offered in the rdkit.Chem.rdMolDescriptors module as shown in the following tutorials.

1. GetRDKitFPGenerator().GetFingerprint() method is actually provides the same functionality as the rdkit.Chem.rdmolops.RDKFingerprint() method. For example:

from rdkit.Chem import AllChem
from rdkit.Chem import rdFingerprintGenerator
from rdkit.DataStructs import cDataStructs
mol = AllChem.MolFromSmiles('CCCC')
gen = rdFingerprintGenerator.GetRDKitFPGenerator(fpSize=64)
fp = gen.GetFingerprint(mol)
display(fp.ToBitString())

fp = AllChem.RDKFingerprint(mol, fpSize=64)
display(fp.ToBitString())

# output:
'0000010100000000000001000000100000000000000110000000000000000000'
'0000010100000000000001000000100000000000000110000000000000000000'

2. GetRDKitFPGenerator().GetSparseCountFingerprint() method is actually provides the same functionality as the rdkit.Chem.rdMolDescriptors.UnfoldedRDKFingerprintCountBased() method, except that GetRDKitFPGenerator() generates 2 identifiers per subgraph by default. But you can reduce it to 1 using the numBitsPerFeature=1 option. For example:

mol = AllChem.MolFromSmiles('CCCC')
gen = rdFingerprintGenerator.GetRDKitFPGenerator(numBitsPerFeature=1) 
fp = gen.GetSparseCountFingerprint(mol)
display(cDataStructs.ULongSparseIntVect.GetNonzeroElements(fp))

fp = AllChem.UnfoldedRDKFingerprintCountBased(mol)
display(cDataStructs.ULongSparseIntVect.GetNonzeroElements(fp))

# output: 
{1173125914: 2, 1244535424: 1, 2245384272: 2, 2246728737: 2, 3542456614: 2}
{1173125914: 2, 1244535424: 1, 2245384272: 2, 2246728737: 2, 3542456614: 2}

3. GetRDKitFPGenerator().GetCountFingerprint() method is actually provides the same functionality as the rdkit.Chem.rdMolDescriptors.RDKFingerprint() method, except that GetCountFingerprint() returns folded identifiers and their duplication counts. Those folded identifiers are ON (value of 1) bits locations in the bit string returned by RDKFingerprint(). For example:

mol = AllChem.MolFromSmiles('CCCC')
gen = rdFingerprintGenerator.GetRDKitFPGenerator(fpSize=64, 
  numBitsPerFeature=1) 
fp = gen.GetCountFingerprint(mol)
display(cDataStructs.UIntSparseIntVect.GetNonzeroElements(fp))

fp = AllChem.RDKFingerprint(mol, fpSize=64, nBitsPerHash=1)
display(fp.ToBitString())

# output: 
{21: 2, 28: 3, 43: 1}
'0000000000000000000001000000100000000000000100000000000000000000'
                      ^      ^              ^
                      21     28             43

4. GetMorganGenerator().GetRDKitFPGenerator() method is actually provides the same functionality as the rdkit.Chem.rdMolDescriptors.UnfoldedRDKFingerprintCountBased() method, except that GetSparseFingerprint() returns identifiers in a SparseBitVect object. For example:

mol = AllChem.MolFromSmiles('CCCC')
gen = rdFingerprintGenerator.GetRDKitFPGenerator(numBitsPerFeature=1) 
fp = gen.GetSparseFingerprint(mol)
display(fp)
display(fp.GetNumBits())
display(fp.GetNumOnBits())
display(list(fp.GetOnBits()))

fp = AllChem.UnfoldedRDKFingerprintCountBased(mol)
display(fp)
display(cDataStructs.ULongSparseIntVect.GetNonzeroElements(fp))

# output:
<rdkit.DataStructs.cDataStructs.SparseBitVect at 0x7fcbd562ac40>
4294967295
3
[-1382976661, -19262180, 1940446997]
<rdkit.DataStructs.cDataStructs.ULongSparseIntVect at 0x7fcbd554ed60>
{1940446997: 2, 2911990635: 1, 4275705116: 3}