Source Code for Module Chem.EState.Fingerprinter

 1  # $Id: Fingerprinter.py 455 2007-12-18 06:37:48Z glandrum $ 
 2  # 
 3  #  Copyright (C) 2002-2006  greg Landrum and Rational Discovery LLC 
 4  # 
 5  #   @@ All Rights Reserved  @@ 
 6  # 
 7  """  EState fingerprinting 
 8   
 9  """ 
10  import Numeric 
11  from Chem.EState import EStateIndices 
12  from Chem.EState import AtomTypes 
13   
14 -def FingerprintMol(mol): 
15    """ generates the EState fingerprints for the molecule 
16   
17    Concept from the paper: Hall and Kier JCICS _35_ 1039-1045 (1995) 
18   
19    two numeric arrays are returned: 
20      The first (of ints) contains the number of times each possible atom type is hit 
21      The second (of floats) contains the sum of the EState indices for atoms of 
22        each type. 
23   
24    """ 
25    if AtomTypes.esPatterns is None: 
26      AtomTypes.BuildPatts() 
27    esIndices = EStateIndices(mol) 
28   
29    nPatts = len(AtomTypes.esPatterns) 
30    counts = Numeric.zeros(nPatts,Numeric.Int) 
31    sums = Numeric.zeros(nPatts,Numeric.Float) 
32   
33    for i,(name,pattern) in enumerate(AtomTypes.esPatterns): 
34      matches = mol.GetSubstructMatches(pattern,uniquify=1) 
35      counts[i] = len(matches) 
36      for match in matches: 
37        sums[i] += esIndices[match[0]] 
38    return counts,sums 
39   
40   
41  if __name__ == '__main__': 
42    import Chem 
43    smis = ['CC','CCC','c1[nH]cnc1CC(N)C(O)=O','NCCc1ccc(O)c(O)c1'] 
44    for smi in smis: 
45      m = Chem.MolFromSmiles(smi) 
46      print smi,Chem.MolToSmiles(m) 
47      types = AtomTypes.TypeAtoms(m) 
48      for i in range(m.GetNumAtoms()): 
49        print '%d %4s: %s'%(i+1,m.GetAtomWithIdx(i).GetSymbol(),str(types[i])) 
50      es = EStateIndices(m) 
51      counts,sums = FingerprintMol(m) 
52      for i in range(len(AtomTypes.esPatterns)): 
53        if counts[i]: 
54          name,patt = AtomTypes.esPatterns[i] 
55          print '%6s, % 2d, % 5.4f'%(name,counts[i],sums[i]) 
56      for i in range(len(es)): 
57        print '% 2d, % 5.4f'%(i+1,es[i]) 
58      print '--------' 
59