Source Code for Module rdkit.Chem.FeatMaps.FeatMapParser

  1  # $Id: FeatMapParser.py 997 2009-02-25 06:12:43Z glandrum $ 
  2  # 
  3  # Copyright (C) 2006 Greg Landrum 
  4  # 
  5  #   @@ All Rights Reserved  @@ 
  6  # 
  7  from rdkit import Geometry 
  8  from rdkit.Chem import ChemicalFeatures 
  9  from rdkit.Chem.FeatMaps import FeatMaps,FeatMapPoint 
 10  import re 
 11   
 12  """ 
 13   
 14  ScoreMode=All 
 15  DirScoreMode=Ignore 
 16   
 17  BeginParams 
 18    family=Aromatic radius=2.5 width=1.0 profile=Gaussian 
 19    family=Acceptor radius=1.5 
 20  EndParams 
 21   
 22  # optional 
 23  BeginPoints 
 24    family=Acceptor pos=(1.0, 0.0, 5.0) weight=1.25 dir=(1, 1, 0) 
 25    family=Aromatic pos=(0.0,1.0,0.0) weight=2.0 dir=(0,0,1) dir=(0,0,-1) 
 26    family=Acceptor pos=(1.0,1.0,2.0) weight=1.25 
 27  EndPoints 
 28   
 29  """ 
 30   
 31 -class FeatMapParseError(ValueError): 
 32    pass 
 33   
 34 -class FeatMapParser(object): 
 35    data=None 
 36 -  def __init__(self,file=None,data=None): 
 37      if file: 
 38        self.data=file.readlines() 
 39      elif data: 
 40        self.SetData(data) 
 41      self._lineNum=0 
 42   
 43 -  def SetData(self,data): 
 44      if isinstance(data,str): 
 45        self.data=data.split('\n') 
 46      else: 
 47        self.data=data 
 48      self._lineNum=0 
 49   
 50 -  def _NextLine(self): 
 51      txt = '' 
 52      while 1: 
 53        try: 
 54          l = self.data[self._lineNum].split('#')[0].strip() 
 55        except IndexError: 
 56          break 
 57        self._lineNum+=1 
 58        if l: 
 59          txt += l 
 60          if l[-1]!='\\': 
 61            break 
 62      return txt 
 63   
 64 -  def Parse(self,featMap=None): 
 65      if featMap is None: 
 66        featMap = FeatMaps.FeatMap() 
 67         
 68      l = self._NextLine().strip() 
 69      while l: 
 70        splitL = l.split('=') 
 71        if len(splitL)==1: 
 72          keyword=splitL[0].strip().lower() 
 73          if keyword=='beginpoints': 
 74            pts=self.ParseFeatPointBlock() 
 75            for pt in pts: 
 76              featMap.AddFeatPoint(pt) 
 77          elif keyword=='beginparams': 
 78            featMap.params=self.ParseParamBlock() 
 79          else: 
 80            raise FeatMapParseError,'Unrecognized keyword %s on line %d'%(keyword,self._lineNum) 
 81        else: 
 82          keyword = splitL[0].strip().lower() 
 83          val = splitL[1].strip() 
 84          if keyword=='scoremode': 
 85            try: 
 86              featMap.scoreMode=getattr(FeatMaps.FeatMapScoreMode,val) 
 87            except AttributeError: 
 88              raise FeatMapParseError,'ScoreMode %s not recognized on line %d'%(val,self._lineNum) 
 89          elif keyword=='dirscoremode': 
 90            try: 
 91              featMap.dirScoreMode=getattr(FeatMaps.FeatDirScoreMode,val) 
 92            except AttributeError: 
 93              raise FeatMapParseError,'DirScoreMode %s not recognized on line %d'%(val,self._lineNum) 
 94          else: 
 95            raise FeatMapParseError,'Unrecognized keyword %s on line %d'%(keyword,self._lineNum) 
 96        l = self._NextLine().strip() 
 97      return featMap 
 98       
 99 -  def ParseParamBlock(self): 
100      paramLineSplitter = re.compile(r'([a-zA-Z]+) *= *(\S+)') 
101      params = {} 
102   
103      l = self._NextLine() 
104      while l and l!='EndParams': 
105        param = FeatMaps.FeatMapParams() 
106        vals=paramLineSplitter.findall(l) 
107        for name,val in vals: 
108          name = name.lower() 
109          if name=='family': 
110            family=val 
111          elif name=='radius': 
112            param.radius=float(val) 
113          elif name=='width': 
114            param.width=float(val) 
115          elif name=='profile': 
116            try: 
117              param.featProfile=getattr(param.FeatProfile,val) 
118            except AttributeError: 
119              raise FeatMapParseError,'Profile %s not recognized on line %d'%(val,self._lineNum) 
120          else: 
121            raise FeatMapParseError,'FeatMapParam option %s not recognized on line %d'%(name,self._lineNum) 
122        params[family]=param 
123        l = self._NextLine() 
124   
125      if l!='EndParams': 
126        raise FeatMapParseError('EndParams line not found') 
127       
128      return params 
129   
130 -  def _parsePoint(self,txt): 
131      txt = txt.strip() 
132      startP=0 
133      endP=len(txt) 
134      if txt[0]=='(': 
135        startP += 1 
136      if txt[-1]==')': 
137        endP -= 1 
138      txt = txt[startP:endP] 
139      splitL = txt.split(',') 
140      if len(splitL) != 3: 
141        raise ValueError,'Bad location string' 
142      vs = [float(x) for x in splitL] 
143      pt = Geometry.Point3D(vs[0],vs[1],vs[2]) 
144      return pt 
145       
146   
147 -  def ParseFeatPointBlock(self): 
148      featLineSplitter = re.compile(r'([a-zA-Z]+) *= *') 
149      feats = [] 
150       
151      l = self._NextLine() 
152      while l and l!='EndPoints': 
153        vals=featLineSplitter.split(l) 
154        while vals.count(''): vals.remove('') 
155        p = FeatMapPoint.FeatMapPoint() 
156   
157        i=0 
158        while i<len(vals): 
159          name = vals[i].lower() 
160          if name=='family': 
161            i+=1 
162            val = vals[i].strip() 
163            p.SetFamily(val) 
164          elif name=='weight': 
165            i+=1 
166            val = float(vals[i]) 
167            p.weight = val 
168          elif name=='pos': 
169            i+=1 
170            val = vals[i] 
171            pos = self._parsePoint(val) 
172            p.SetPos(pos) 
173          elif name=='dir': 
174            i+=1 
175            val = vals[i] 
176            pos = self._parsePoint(val) 
177            p.featDirs.append(pos) 
178          else: 
179            raise FeatMapParseError,'FeatPoint option %s not recognized on line %d'%(name,self._lineNum) 
180          i+=1 
181        feats.append(p) 
182        l = self._NextLine() 
183      return feats 
184  #------------------------------------ 
185  # 
186  #  doctest boilerplate 
187  # 
188 -def _test(): 
189    import doctest,sys 
190    return doctest.testmod(sys.modules["__main__"]) 
191   
192  if __name__ == '__main__': 
193    import sys 
194    failed,tried = _test() 
195    sys.exit(failed) 
196