Source Code for Module rdkit.Chem.MolDb.FingerprintUtils

  1  # $Id: FingerprintUtils.py 1647 2011-02-24 08:00:39Z glandrum $ 
  2  # 
  3  # Copyright (C) 2009 Greg Landrum 
  4  #  All Rights Reserved 
  5  # 
  6  import cPickle 
  7  from rdkit import DataStructs,Chem 
  8  from rdkit import Chem 
  9   
 10  similarityMethods={'RDK':DataStructs.ExplicitBitVect, 
 11                     'AtomPairs':DataStructs.IntSparseIntVect, 
 12                     'TopologicalTorsions':DataStructs.LongSparseIntVect, 
 13                     'Pharm2D':DataStructs.SparseBitVect, 
 14                     'Gobbi2D':DataStructs.SparseBitVect, 
 15                     'Morgan':DataStructs.UIntSparseIntVect 
 16                     } 
 17  supportedSimilarityMethods=similarityMethods.keys() 
 18   
 19   
 20 -class LayeredOptions: 
 21    loadLayerFlags=0xFFFFFFFF 
 22    searchLayerFlags=0x7 
 23    minPath=1 
 24    maxPath=6 
 25    fpSize=1024 
 26    wordSize=32 
 27    nWords=fpSize//wordSize 
 28    @staticmethod 
 29 -  def GetFingerprint(mol,query=True): 
 30      if query: 
 31        flags=LayeredOptions.searchLayerFlags 
 32      else: 
 33        flags=LayeredOptions.loadLayerFlags 
 34      return Chem.LayeredFingerprint(mol,layerFlags=flags, 
 35                                     minPath=LayeredOptions.minPath,maxPath=LayeredOptions.maxPath, 
 36                                     fpSize=LayeredOptions.fpSize) 
 37    @staticmethod 
 38 -  def GetWords(mol,query=True): 
 39      txt = LayeredOptions.GetFingerprint(mol,query=query).ToBitString() 
 40      words = [int(txt[x:x+32],2) for x in range(0,len(txt),32)] 
 41      return  words 
 42   
 43    @staticmethod 
 44 -  def GetQueryText(mol,query=True): 
 45      words = LayeredOptions.GetWords(mol,query=query) 
 46      colqs = [] 
 47      for idx,word in enumerate(words): 
 48        if not word: 
 49          continue 
 50        idx = idx+1 
 51        colqs.append('%(word)d&Col_%(idx)d=%(word)d'%locals()) 
 52      return  ' and '.join(colqs) 
 53   
 54   
 55   
 56 -def BuildSigFactory(options=None,fdefFile=None, 
 57                      bins=[(2,3),(3,4),(4,5),(5,6),(6,7),(7,8),(8,100)], 
 58                      skipFeats=('LumpedHydrophobe','ZnBinder')): 
 59    if options: 
 60      fdefFile = options.fdefFile 
 61    if not fdefFile: 
 62      raise ValueError,'bad fdef file' 
 63    from rdkit.Chem import ChemicalFeatures 
 64    from rdkit.Chem.Pharm2D import SigFactory 
 65    featFactory = ChemicalFeatures.BuildFeatureFactory(fdefFile) 
 66    sigFactory = SigFactory.SigFactory(featFactory, 
 67                                       skipFeats=skipFeats, 
 68                                       trianglePruneBins=False) 
 69    sigFactory.SetBins(bins) 
 70    return sigFactory 
 71   
 72 -def BuildAtomPairFP(mol): 
 73    from rdkit.Chem.AtomPairs import Pairs 
 74    fp=Pairs.GetAtomPairFingerprintAsIntVect(mol) 
 75    fp._sumCache = fp.GetTotalVal() 
 76    return fp 
 77 -def BuildTorsionsFP(mol): 
 78    from rdkit.Chem.AtomPairs import Torsions 
 79    fp=Torsions.GetTopologicalTorsionFingerprintAsIntVect(mol) 
 80    fp._sumCache = fp.GetTotalVal() 
 81    return fp 
 82 -def BuildRDKitFP(mol): 
 83    fp=Chem.RDKFingerprint(mol,nBitsPerHash=1) 
 84    return fp 
 85 -def BuildPharm2DFP(mol): 
 86    global sigFactory 
 87    from rdkit.Chem.Pharm2D import Generate 
 88    try: 
 89      fp=Generate.Gen2DFingerprint(mol,sigFactory) 
 90    except IndexError: 
 91      print 'FAIL:',Chem.MolToSmiles(mol,True) 
 92      raise 
 93    return fp 
 94 -def BuildMorganFP(mol): 
 95    from rdkit.Chem import rdMolDescriptors 
 96    fp = rdMolDescriptors.GetMorganFingerprint(mol,2) 
 97    fp._sumCache = fp.GetTotalVal() 
 98    return fp 
 99   
100 -def DepickleFP(pkl,similarityMethod): 
101      try: 
102          klass=similarityMethods[similarityMethod] 
103          fp = klass(pkl) 
104      except: 
105          import traceback 
106          traceback.print_exc() 
107          fp = cPickle.loads(pkl) 
108      return fp 
109