Source Code for Module rdkit.Chem.AtomPairs.Pairs

  1  # $Id: Pairs.py 997 2009-02-25 06:12:43Z glandrum $ 
  2  # 
  3  #  Copyright (C) 2004-2007 Greg Landrum and Rational Discovery LLC 
  4  # 
  5  #   @@ All Rights Reserved  @@ 
  6  # 
  7  """ Contains an implementation of Atom-pair fingerprints, as 
  8  described in: 
  9   
 10  R.E. Carhart, D.H. Smith, R. Venkataraghavan; 
 11  "Atom Pairs as Molecular Features in Structure-Activity Studies: 
 12  Definition and Applications" JCICS 25, 64-73 (1985). 
 13   
 14  """ 
 15  from rdkit.DataStructs import IntSparseIntVect 
 16  from rdkit import Chem 
 17  from rdkit.Chem import rdMolDescriptors 
 18  from rdkit.Chem.AtomPairs import Utils 
 19  from rdkit import DataStructs 
 20   
 21  GetAtomPairFingerprint=rdMolDescriptors.GetAtomPairFingerprint 
 22  GetAtomPairFingerprintAsIntVect=rdMolDescriptors.GetAtomPairFingerprint 
 23   
 24  numPathBits=rdMolDescriptors.AtomPairsParameters.numPathBits 
 25  _maxPathLen=(1<<numPathBits)-1 
 26  numFpBits=numPathBits+2*rdMolDescriptors.AtomPairsParameters.codeSize 
 27  fpLen=1L<<numFpBits 
 28   
 29 -def pyScorePair(at1,at2,dist,atomCodes=None): 
 30    """ Returns a score for an individual atom pair. 
 31   
 32    >>> m = Chem.MolFromSmiles('CCCCC') 
 33    >>> c1 = Utils.GetAtomCode(m.GetAtomWithIdx(0)) 
 34    >>> c2 = Utils.GetAtomCode(m.GetAtomWithIdx(1)) 
 35    >>> c3 = Utils.GetAtomCode(m.GetAtomWithIdx(2)) 
 36    >>> t = 1 | min(c1,c2)<<numPathBits | max(c1,c2)<<(rdMolDescriptors.AtomPairsParameters.codeSize+numPathBits) 
 37    >>> pyScorePair(m.GetAtomWithIdx(0),m.GetAtomWithIdx(1),1)==t 
 38    1 
 39    >>> pyScorePair(m.GetAtomWithIdx(1),m.GetAtomWithIdx(0),1)==t 
 40    1 
 41    >>> t = 2 | min(c1,c3)<<numPathBits | max(c1,c3)<<(rdMolDescriptors.AtomPairsParameters.codeSize+numPathBits) 
 42    >>> pyScorePair(m.GetAtomWithIdx(0),m.GetAtomWithIdx(2),2)==t 
 43    1 
 44    >>> pyScorePair(m.GetAtomWithIdx(0),m.GetAtomWithIdx(2),2, 
 45    ...  atomCodes=(Utils.GetAtomCode(m.GetAtomWithIdx(0)),Utils.GetAtomCode(m.GetAtomWithIdx(2))))==t 
 46    1 
 47   
 48    """ 
 49    if not atomCodes: 
 50      code1 = Utils.GetAtomCode(at1) 
 51      code2 = Utils.GetAtomCode(at2) 
 52    else: 
 53      code1,code2=atomCodes 
 54    accum = int(dist) % _maxPathLen 
 55    accum |= min(code1,code2) << numPathBits 
 56    accum |= max(code1,code2) << (rdMolDescriptors.AtomPairsParameters.codeSize+numPathBits) 
 57    return accum 
 58   
 59 -def ExplainPairScore(score): 
 60    """  
 61    >>> m = Chem.MolFromSmiles('C=CC') 
 62    >>> score = pyScorePair(m.GetAtomWithIdx(0),m.GetAtomWithIdx(1),1) 
 63    >>> ExplainPairScore(score) 
 64    (('C', 1, 1), 1, ('C', 2, 1)) 
 65    >>> score = pyScorePair(m.GetAtomWithIdx(0),m.GetAtomWithIdx(2),2) 
 66    >>> ExplainPairScore(score) 
 67    (('C', 1, 0), 2, ('C', 1, 1)) 
 68    >>> score = pyScorePair(m.GetAtomWithIdx(1),m.GetAtomWithIdx(2),1) 
 69    >>> ExplainPairScore(score) 
 70    (('C', 1, 0), 1, ('C', 2, 1)) 
 71    >>> score = pyScorePair(m.GetAtomWithIdx(2),m.GetAtomWithIdx(1),1) 
 72    >>> ExplainPairScore(score) 
 73    (('C', 1, 0), 1, ('C', 2, 1)) 
 74   
 75    """ 
 76    codeMask = (1<<rdMolDescriptors.AtomPairsParameters.codeSize)-1 
 77    pathMask = (1<<numPathBits)-1 
 78    dist = score&pathMask 
 79   
 80    score = score>>numPathBits 
 81    code1 = score&codeMask 
 82    score = score>>rdMolDescriptors.AtomPairsParameters.codeSize 
 83    code2 = score&codeMask 
 84   
 85    res = Utils.ExplainAtomCode(code1),dist,Utils.ExplainAtomCode(code2) 
 86    return res 
 87   
 88 -def GetAtomPairFingerprintAsBitVect(mol): 
 89    """ Returns the Atom-pair fingerprint for a molecule as 
 90    a SparseBitVect. Note that this doesn't match the standard 
 91    definition of atom pairs, which uses counts of the 
 92    pairs, not just their presence. 
 93   
 94    **Arguments**: 
 95   
 96      - mol: a molecule 
 97   
 98    **Returns**: a SparseBitVect 
 99   
100    >>> m = Chem.MolFromSmiles('CCC') 
101    >>> v = [ pyScorePair(m.GetAtomWithIdx(0),m.GetAtomWithIdx(1),1), 
102    ...       pyScorePair(m.GetAtomWithIdx(0),m.GetAtomWithIdx(2),2), 
103    ...     ] 
104    >>> v.sort() 
105    >>> fp = GetAtomPairFingerprintAsBitVect(m) 
106    >>> list(fp.GetOnBits())==v 
107    True 
108     
109    """ 
110    res = DataStructs.SparseBitVect(fpLen) 
111    fp = rdMolDescriptors.GetAtomPairFingerprint(mol) 
112    for val in fp.GetNonzeroElements().keys(): 
113      res.SetBit(val) 
114    return res 
115   
116  #------------------------------------ 
117  # 
118  #  doctest boilerplate 
119  # 
120 -def _test(): 
121    import doctest,sys 
122    return doctest.testmod(sys.modules["__main__"]) 
123   
124   
125  if __name__ == '__main__': 
126    import sys 
127    failed,tried = _test() 
128    sys.exit(failed) 
129