Source Code for Module rdkit.Chem.ChemUtils.BulkTester

 1  # $Id: BulkTester.py 997 2009-02-25 06:12:43Z glandrum $ 
 2  # 
 3  #  Copyright (C) 2005-2006 Rational Discovery LLC 
 4  # 
 5  #   @@ All Rights Reserved  @@ 
 6  # 
 7  from rdkit import Chem 
 8  import sys 
 9  from rdkit.Chem import Randomize 
10   
11 -def TestMolecule(mol): 
12    try: 
13      Chem.SanitizeMol(mol) 
14      mol = Chem.RemoveHs(mol) 
15    except ValueError,msg: 
16      return -1 
17    except: 
18      import traceback 
19      traceback.print_exc() 
20      return -2 
21    if mol.GetNumAtoms(): 
22      try: 
23        Randomize.CheckCanonicalization(mol,10) 
24      except: 
25        import traceback 
26        traceback.print_exc() 
27        return -3 
28    return 0 
29       
30       
31 -def TestSupplier(suppl,stopAfter=-1,reportInterval=100,reportTo=sys.stderr, 
32                   nameProp='_Name'): 
33    nDone = 0 
34    nFailed = 0 
35    while 1: 
36      try: 
37        mol = suppl.next() 
38      except StopIteration: 
39        break 
40      except: 
41        import traceback 
42        traceback.print_exc() 
43        nFailed += 1 
44        reportTo.flush() 
45        print >>reportTo,'Failure at mol %d'%nDone 
46      else: 
47        if mol: 
48          ok = TestMolecule(mol) 
49        else: 
50          ok = -3 
51        if ok<0: 
52          nFailed += 1 
53          reportTo.flush() 
54          if ok==-3: 
55            print >>reportTo,'Canonicalization', 
56          print >>reportTo,'Failure at mol %d'%nDone, 
57          if mol: 
58            print >>reportTo,mol.GetProp(nameProp), 
59          print >>reportTo,'' 
60   
61             
62      nDone += 1 
63      if nDone==stopAfter: 
64        break 
65      if not nDone%reportInterval: 
66        print 'Done %d molecules, %d failures'%(nDone,nFailed) 
67    return nDone,nFailed 
68  if __name__=='__main__': 
69    suppl = Chem.SDMolSupplier(sys.argv[1],False) 
70    if len(sys.argv)>2: 
71      nameProp = sys.argv[2] 
72    else: 
73      nameProp = '_Name' 
74       
75    nDone,nFailed = TestSupplier(suppl,nameProp=nameProp) 
76    print '%d failures in %d mols'%(nFailed,nDone) 
77