Source Code for Module rdkit.Chem.MolDb.FingerprintUtils

  1  # $Id$ 
  2  # 
  3  # Copyright (C) 2009 Greg Landrum 
  4  #  All Rights Reserved 
  5  # 
  6  from __future__ import print_function 
  7  from rdkit.six.moves import cPickle 
  8  from rdkit.six import iterkeys 
  9  from rdkit import DataStructs,Chem 
 10  from rdkit import Chem 
 11   
 12  similarityMethods={'RDK':DataStructs.ExplicitBitVect, 
 13                     'AtomPairs':DataStructs.IntSparseIntVect, 
 14                     'TopologicalTorsions':DataStructs.LongSparseIntVect, 
 15                     'Pharm2D':DataStructs.SparseBitVect, 
 16                     'Gobbi2D':DataStructs.SparseBitVect, 
 17                     'Morgan':DataStructs.UIntSparseIntVect 
 18                     } 
 19  supportedSimilarityMethods=list(iterkeys(similarityMethods)) 
 20   
 21   
 22 -class LayeredOptions: 
 23    loadLayerFlags=0xFFFFFFFF 
 24    searchLayerFlags=0x7 
 25    minPath=1 
 26    maxPath=6 
 27    fpSize=1024 
 28    wordSize=32 
 29    nWords=fpSize//wordSize 
 30    @staticmethod 
 31 -  def GetFingerprint(mol,query=True): 
 32      if query: 
 33        flags=LayeredOptions.searchLayerFlags 
 34      else: 
 35        flags=LayeredOptions.loadLayerFlags 
 36      return Chem.LayeredFingerprint(mol,layerFlags=flags, 
 37                                     minPath=LayeredOptions.minPath,maxPath=LayeredOptions.maxPath, 
 38                                     fpSize=LayeredOptions.fpSize) 
 39    @staticmethod 
 40 -  def GetWords(mol,query=True): 
 41      txt = LayeredOptions.GetFingerprint(mol,query=query).ToBitString() 
 42      words = [int(txt[x:x+32],2) for x in range(0,len(txt),32)] 
 43      return  words 
 44   
 45    @staticmethod 
 46 -  def GetQueryText(mol,query=True): 
 47      words = LayeredOptions.GetWords(mol,query=query) 
 48      colqs = [] 
 49      for idx,word in enumerate(words): 
 50        if not word: 
 51          continue 
 52        idx = idx+1 
 53        colqs.append('%(word)d&Col_%(idx)d=%(word)d'%locals()) 
 54      return  ' and '.join(colqs) 
 55   
 56   
 57   
 58 -def BuildSigFactory(options=None,fdefFile=None, 
 59                      bins=[(2,3),(3,4),(4,5),(5,6),(6,7),(7,8),(8,100)], 
 60                      skipFeats=('LumpedHydrophobe','ZnBinder')): 
 61    if options: 
 62      fdefFile = options.fdefFile 
 63    if not fdefFile: 
 64      raise ValueError('bad fdef file') 
 65    from rdkit.Chem import ChemicalFeatures 
 66    from rdkit.Chem.Pharm2D import SigFactory 
 67    featFactory = ChemicalFeatures.BuildFeatureFactory(fdefFile) 
 68    sigFactory = SigFactory.SigFactory(featFactory, 
 69                                       skipFeats=skipFeats, 
 70                                       trianglePruneBins=False) 
 71    sigFactory.SetBins(bins) 
 72    return sigFactory 
 73   
 74 -def BuildAtomPairFP(mol): 
 75    from rdkit.Chem.AtomPairs import Pairs 
 76    fp=Pairs.GetAtomPairFingerprintAsIntVect(mol) 
 77    fp._sumCache = fp.GetTotalVal() 
 78    return fp 
 79 -def BuildTorsionsFP(mol): 
 80    from rdkit.Chem.AtomPairs import Torsions 
 81    fp=Torsions.GetTopologicalTorsionFingerprintAsIntVect(mol) 
 82    fp._sumCache = fp.GetTotalVal() 
 83    return fp 
 84 -def BuildRDKitFP(mol): 
 85    fp=Chem.RDKFingerprint(mol,nBitsPerHash=1) 
 86    return fp 
 87 -def BuildPharm2DFP(mol): 
 88    global sigFactory 
 89    from rdkit.Chem.Pharm2D import Generate 
 90    try: 
 91      fp=Generate.Gen2DFingerprint(mol,sigFactory) 
 92    except IndexError: 
 93      print('FAIL:',Chem.MolToSmiles(mol,True)) 
 94      raise 
 95    return fp 
 96 -def BuildMorganFP(mol): 
 97    from rdkit.Chem import rdMolDescriptors 
 98    fp = rdMolDescriptors.GetMorganFingerprint(mol,2) 
 99    fp._sumCache = fp.GetTotalVal() 
100    return fp 
101   
102 -def DepickleFP(pkl,similarityMethod): 
103      if not isinstance(pkl,(bytes,str)): 
104        pkl = str(pkl) 
105      try: 
106          klass=similarityMethods[similarityMethod] 
107          fp = klass(pkl) 
108      except Exception: 
109          import traceback 
110          traceback.print_exc() 
111          fp = cPickle.loads(pkl) 
112      return fp 
113