Merge pull request #14 from Jnelen/master

Readded support for the signature fingerprint

Author: hcji

Example/test_fingerprint.py

@@ -8,7 +8,7 @@
 import numpy as np
 from PyFingerprint.fingerprint import get_fingerprint, get_fingerprints
 
-cdktypes = ['standard', 'extended', 'graph', 'maccs', 'pubchem', 'estate', 'hybridization', 'lingo', 'klekota-roth', 'shortestpath', 'cdk-substructure', 'circular', 'atompairs']
+cdktypes = ['standard', 'extended', 'graph', 'maccs', 'pubchem', 'estate', 'hybridization', 'lingo', 'klekota-roth', 'shortestpath', 'cdk-substructure', 'circular', 'atompairs', 'signature']
 rdktypes = ['rdkit', 'morgan', 'rdk-maccs', 'topological-torsion', 'avalon', 'atom-pair', 'rdk-descriptor']
 babeltypes = ['fp2', 'fp3', 'fp4', 'spectrophore']
 vectypes = ['mol2vec', 'heteroencoder']

PyFingerprint/cdk.py

@@ -91,22 +91,10 @@ def cdk_fingerprint(smi, fp_type="standard", size=1024, depth=None):
     elif fp_type == 'pubchem':
         nbit = 881
     elif fp_type == 'klekota-roth':
-        nbit = 4860      
-    elif fp_type == 'signature':
-        nbit = None
-        print("Signature_FP")
-        fingerprinter = cdk.fingerprint.SignatureFingerprinter()
-        mol = cdk_parser_smiles(smi)
-        print(fingerprinter.getSize())
-        print(fingerprinter.getBitFingerprint(mol).getSetbits())
-        print(fingerprinter.getBitFingerprint(mol).size())
-        print(fingerprinter.getRawFingerprint(mol))
-        
+        nbit = 4860
     else:
         nbit = size
 
-    
-
     # Pull from cache if it exists
     if (fp_type, size, depth) in fp_map: 
         fingerprinter = fp_map[(fp_type, size, depth)]
@@ -116,4 +104,6 @@ def cdk_fingerprint(smi, fp_type="standard", size=1024, depth=None):
 
     fp_obj = fingerprinter.getBitFingerprint(mol)
     bits = list(fp_obj.getSetbits())
+    if fp_type == "signature":
+        bits = list(set((bit%nbit) for bit in bits))
     return bits, nbit

PyFingerprint/fingerprint.py

@@ -22,7 +22,7 @@
     hc_enable = False
 
 
-cdktypes = ['standard', 'extended', 'graph', 'maccs', 'pubchem', 'estate', 'hybridization', 'lingo', 'klekota-roth', 'shortestpath', 'cdk-substructure', 'circular', 'cdk-atompairs']
+cdktypes = ['standard', 'extended', 'graph', 'maccs', 'pubchem', 'estate', 'hybridization', 'lingo', 'klekota-roth', 'shortestpath', 'cdk-substructure', 'circular', 'cdk-atompairs', 'signature']
 rdktypes = ['rdkit', 'morgan', 'rdk-maccs', 'topological-torsion', 'avalon', 'atom-pair', 'rdk-descriptor']
 babeltypes = ['fp2', 'fp3', 'fp4', 'spectrophore']
 vectypes = ['mol2vec', 'heteroencoder']