Source Code for Module rdkit.Chem.AtomPairs.Torsions

  1  # $Id: Torsions.py 997 2009-02-25 06:12:43Z glandrum $ 
  2  # 
  3  #  Copyright (C) 2004-2006 Rational Discovery LLC 
  4  # 
  5  #   @@ All Rights Reserved  @@ 
  6  # 
  7  """ Contains an implementation of Topological-torsion fingerprints, as 
  8  described in: 
  9   
 10  R. Nilakantan, N. Bauman, J. S. Dixon, R. Venkataraghavan; 
 11  "Topological Torsion: A New Molecular Descriptor for SAR Applications. 
 12  Comparison with Other Descriptors" JCICS 27, 82-85 (1987). 
 13   
 14  """ 
 15  from rdkit import Chem 
 16  from rdkit.Chem import rdMolDescriptors 
 17  import Utils 
 18   
 19 -def pyScorePath(mol,path,size,atomCodes=None): 
 20    """ Returns a score for an individual path. 
 21   
 22    >>> m = Chem.MolFromSmiles('CCCCC') 
 23    >>> c1 = long(Utils.GetAtomCode(m.GetAtomWithIdx(0),1)) 
 24    >>> c2 = long(Utils.GetAtomCode(m.GetAtomWithIdx(1),2)) 
 25    >>> c3 = long(Utils.GetAtomCode(m.GetAtomWithIdx(2),2)) 
 26    >>> c4 = long(Utils.GetAtomCode(m.GetAtomWithIdx(3),1)) 
 27    >>> t = c1 | (c2 << rdMolDescriptors.AtomPairsParameters.codeSize) | (c3 << (rdMolDescriptors.AtomPairsParameters.codeSize*2)) | (c4 << (rdMolDescriptors.AtomPairsParameters.codeSize*3)) 
 28    >>> pyScorePath(m,(0,1,2,3),4)==t 
 29    1 
 30   
 31    The scores are path direction independent: 
 32    >>> pyScorePath(m,(3,2,1,0),4)==t 
 33    1 
 34   
 35   
 36    >>> m = Chem.MolFromSmiles('C=CC(=O)O') 
 37    >>> c1 = long(Utils.GetAtomCode(m.GetAtomWithIdx(0),1)) 
 38    >>> c2 = long(Utils.GetAtomCode(m.GetAtomWithIdx(1),2)) 
 39    >>> c3 = long(Utils.GetAtomCode(m.GetAtomWithIdx(2),2)) 
 40    >>> c4 = long(Utils.GetAtomCode(m.GetAtomWithIdx(4),1)) 
 41    >>> t = c1 | (c2 << rdMolDescriptors.AtomPairsParameters.codeSize) | (c3 << (rdMolDescriptors.AtomPairsParameters.codeSize*2)) | (c4 << (rdMolDescriptors.AtomPairsParameters.codeSize*3)) 
 42    >>> pyScorePath(m,(0,1,2,4),4)==t 
 43    1 
 44   
 45    """ 
 46    codes = [None]*size 
 47    for i in range(size): 
 48      if i==0 or i==(size-1): 
 49        sub = 1 
 50      else: 
 51        sub = 2 
 52      if not atomCodes: 
 53        codes[i] = Utils.GetAtomCode(mol.GetAtomWithIdx(path[i]),sub) 
 54      else: 
 55        base = atomCodes[path[i]] 
 56        codes[i]=base-sub 
 57   
 58    # "canonicalize" the code vector: 
 59    beg=0 
 60    end = len(codes)-1 
 61    while(beg < end): 
 62      if codes[beg] > codes[end]: 
 63        codes.reverse() 
 64        break 
 65      elif codes[beg]==codes[end]: 
 66        beg += 1 
 67        end -= 1 
 68      else: 
 69        break 
 70    accum = 0L 
 71    for i in range(size): 
 72      accum |= long(codes[i]) << (rdMolDescriptors.AtomPairsParameters.codeSize*i) 
 73    return accum 
 74   
 75 -def ExplainPathScore(score,size=4): 
 76    """  
 77   
 78    >>> m = Chem.MolFromSmiles('C=CC') 
 79    >>> score=pyScorePath(m,(0,1,2),3) 
 80    >>> ExplainPathScore(score,3) 
 81    (('C', 1, 0), ('C', 2, 1), ('C', 1, 1)) 
 82   
 83    Again, it's order independent: 
 84    >>> score=pyScorePath(m,(2,1,0),3) 
 85    >>> ExplainPathScore(score,3) 
 86    (('C', 1, 0), ('C', 2, 1), ('C', 1, 1)) 
 87   
 88   
 89    >>> m = Chem.MolFromSmiles('C=CO') 
 90    >>> score=pyScorePath(m,(0,1,2),3) 
 91    >>> ExplainPathScore(score,3) 
 92    (('C', 1, 1), ('C', 2, 1), ('O', 1, 0)) 
 93   
 94    >>> m = Chem.MolFromSmiles('OC=CO') 
 95    >>> score=pyScorePath(m,(0,1,2,3),4) 
 96    >>> ExplainPathScore(score,4) 
 97    (('O', 1, 0), ('C', 2, 1), ('C', 2, 1), ('O', 1, 0)) 
 98   
 99    >>> m = Chem.MolFromSmiles('CC=CO') 
100    >>> score=pyScorePath(m,(0,1,2,3),4) 
101    >>> ExplainPathScore(score,4) 
102    (('C', 1, 0), ('C', 2, 1), ('C', 2, 1), ('O', 1, 0)) 
103   
104   
105    >>> m = Chem.MolFromSmiles('C=CC(=O)O') 
106    >>> score=pyScorePath(m,(0,1,2,3),4) 
107    >>> ExplainPathScore(score,4) 
108    (('C', 1, 1), ('C', 2, 1), ('C', 3, 1), ('O', 1, 1)) 
109    >>> score=pyScorePath(m,(0,1,2,4),4) 
110    >>> ExplainPathScore(score,4) 
111    (('C', 1, 1), ('C', 2, 1), ('C', 3, 1), ('O', 1, 0)) 
112   
113   
114    >>> m = Chem.MolFromSmiles('OOOO') 
115    >>> score=pyScorePath(m,(0,1,2),3) 
116    >>> ExplainPathScore(score,3) 
117    (('O', 1, 0), ('O', 2, 0), ('O', 2, 0)) 
118    >>> score=pyScorePath(m,(0,1,2,3),4) 
119    >>> ExplainPathScore(score,4) 
120    (('O', 1, 0), ('O', 2, 0), ('O', 2, 0), ('O', 1, 0)) 
121   
122    """ 
123    codeMask=(1<<rdMolDescriptors.AtomPairsParameters.codeSize)-1 
124    res=[None]*size 
125    #print '>>>>>>>>>>>',score,size,codeMask 
126    for i in range(size): 
127      if i==0 or i==(size-1): 
128        sub = 1 
129      else: 
130        sub = 2 
131      code = score&codeMask 
132      #print i,code,score 
133      score = score>>rdMolDescriptors.AtomPairsParameters.codeSize 
134      symb,nBranch,nPi = Utils.ExplainAtomCode(code) 
135      expl = symb,nBranch+sub,nPi 
136      res[i] = expl 
137    return tuple(res) 
138   
139  GetTopologicalTorsionFingerprintAsIntVect=rdMolDescriptors.GetTopologicalTorsionFingerprint 
140  GetTopologicalTorsionFingerprint=rdMolDescriptors.GetTopologicalTorsionFingerprint 
141 -def GetTopologicalTorsionFingerprintAsIds(mol,targetSize=4): 
142    iv = rdMolDescriptors.GetTopologicalTorsionFingerprint(mol,targetSize) 
143    res=[] 
144    for k,v in iv.GetNonzeroElements().iteritems(): 
145      res.extend([k]*v) 
146    res.sort() 
147    return res 
148   
149     
150     
151   
152  #------------------------------------ 
153  # 
154  #  doctest boilerplate 
155  # 
156 -def _test(): 
157    import doctest,sys 
158    return doctest.testmod(sys.modules["__main__"]) 
159   
160   
161  if __name__ == '__main__': 
162    import sys 
163    failed,tried = _test() 
164    sys.exit(failed) 
165