def test_RDKit(self):
from cinfony import rdk
mol = rdk.readstring("smi", "CCC")
desc = mol.calcdesc(rdk.descs[0:3])
for d in desc:
if desc[d] != desc[d]:
desc[d] = '?'
expectedDesc = {'fr_Ar_COO': 0, 'Chi4v': 0.0, 'fr_C_O_noCOO': 0}
self.assertEqual(desc,expectedDesc)
def test_RDKit(self):
from cinfony import rdk
mol = rdk.readstring("smi", "CCC")
desc = mol.calcdesc(rdk.descs[0:3])
for d in desc:
if desc[d] != desc[d]:
desc[d] = '?'
expectedDesc = {'fr_Ar_COO': 0, 'Chi4v': 0.0, 'fr_C_O_noCOO': 0}
self.assertEqual(desc,expectedDesc)
def validate(fn, typ):
ofn = fn.replace('.sdf.gz', '_sanitized.sdf')
if not os.path.isfile(ofn):
good = 0
total = 0
o = rdk.Outputfile('sdf', ofn, overwrite=True)
n = 0
for mol in pybel.readfile('sdf', fn):
mol.title = typ + "_" + mol.title + "_" + str(n)
total += 1
n += 1
try:
o.write(rdk.readstring('mol', mol.write('mol')))
good += 1
except Exception, e:
print e
pass
print "%s %% good mols (%s/%s)" % ((good * 100. / total), good, total)
o.close()
def clean(fn):
ligands = os.path.join(os.path.dirname(fn), "..", "PDB",
"ligands_allgood.sdf")
if not os.path.isfile(ligands):
return
inchikeys = set(
[mol.write('inchikey') for mol in pybel.readfile('sdf', ligands)])
ofn = fn.replace('.sdf', '_filtered.sdf')
if not os.path.isfile(ofn):
good = 0
total = 0
o = rdk.Outputfile('sdf', ofn, overwrite=True)
for mol in pybel.readfile('sdf', fn):
total += 1
if mol.write('inchikey') not in inchikeys:
try:
o.write(rdk.readstring('mol', mol.write('mol')))
good += 1
except Exception, e:
#print e
pass
o.close()
print "%s %% remaining mols (%s/%s)" % (
(good * 100. / total), good, total)
def getRdkDescResult(data,descList, radius = 1):
""" Calculates the descriptors for the descList using RDK
It expects an attribute containing smiles with a name defined in AZOrangeConfig.SMILESNAMES
It returns a dataset with the same smiles input variable, and as many variables as the descriptors
returned by the toolkit
"""
if "rdk" not in toolkitsEnabled:
return None
FingerPrints = False
smilesName = getSMILESAttr(data)
if not smilesName: return None
myDescList = [desc.replace(rdkTag,"") for desc in descList if rdkTag in desc]
if not myDescList: return None
if "FingerPrints" in myDescList:
FingerPrints = True
myDescList.remove("FingerPrints")
#Get fingerprints in advance
fingerPrintsAttrs = []
fingerPrintsRes = {}
if FingerPrints:
for ex in data:
mol = str(ex[smilesName].value)
try:
chemMol = rdk.Chem.MolFromSmiles(mol,True)
if not chemMol:
chemMol = rdk.Chem.MolFromSmiles(mol,False)
fingerPrint = rdk.AllChem.GetMorganFingerprint(chemMol,radius)
resDict = fingerPrint.GetNonzeroElements()
except:
continue
fingerPrintsRes[mol] = {}
for ID in resDict:
count = resDict[ID]
name = rdkTag+"FP_"+str(ID)
if name not in [x.name for x in fingerPrintsAttrs]:
fingerPrintsAttrs.append(orange.FloatVariable(name))
fingerPrintsRes[mol][name]=int(count)
resData = orange.ExampleTable(orange.Domain([data.domain[smilesName]] + [orange.FloatVariable(rdkTag+name) for name in myDescList] + [name for name in fingerPrintsAttrs],0))
badCompounds = 0
for ex in data:
newEx = orange.Example(resData.domain)
newEx[smilesName] = ex[smilesName]
molStr = str(newEx[smilesName].value)
# OBS - add something keeping count on the number of unused smiles
try:
chemMol = rdk.Chem.MolFromSmiles(molStr,True)
if not chemMol:
chemMol = rdk.Chem.MolFromSmiles(molStr,False)
mol = rdk.readstring("mol", rdk.Chem.MolToMolBlock(chemMol))
#mol = rdk.readstring("smi", molStr)
moldesc = mol.calcdesc(myDescList)
for desc in myDescList:
newEx[rdkTag+desc] = moldesc[desc]
#Process fingerprints
if FingerPrints:
for desc in fingerPrintsAttrs:
if desc.name in fingerPrintsRes[molStr]:
newEx[desc.name] = fingerPrintsRes[molStr][desc.name]
else:
newEx[desc.name] = 0
resData.append(newEx)
except:
badCompounds += 1
print "Compounds in original data: ",len(data)
print "Compounds able to calculate descs:",len(resData)
print "Ignored Compounds: ",badCompounds
return resData
def get_RMSD_value(refmol, probemol):
"""Input is 2 mol files."""
rdref = rdk.readstring('mol', str(refmol))
rdprobe = rdk.readstring('mol', str(probemol))
return rdk.Chem.AllChem.GetBestRMS(rdref.Mol, rdprobe.Mol)
def getRdkDescResult(data,descList, radius = 1):
""" Calculates the descriptors for the descList using RDK
It expects an attribute containing smiles with a name defined in AZOrangeConfig.SMILESNAMES
It returns a dataset with the same smiles input variable, and as many variables as the descriptors
returned by the toolkit
"""
if "rdk" not in toolkitsEnabled:
return None
FingerPrints = False
smilesName = getSMILESAttr(data)
if not smilesName: return None
FP_desc = []
myDescList = [desc.replace(toolkitsDef["rdk"]["tag"],"") for desc in descList if toolkitsDef["rdk"]["tag"] in desc]
if not myDescList: return None
if "FingerPrints" in myDescList:
FingerPrints = True
myDescList.remove("FingerPrints")
if sum(["FP_" in fp for fp in myDescList]):
tmpDescList = []
FingerPrints = True
for attr in myDescList:
if "FP_" not in attr:
tmpDescList.append(attr)
else:
FP_desc.append(attr)
myDescList = tmpDescList
#Get fingerprints in advance
fingerPrintsAttrs = []
fingerPrintsRes = {}
if FingerPrints:
for ex in data:
mol = str(ex[smilesName].value)
try:
chemMol = rdk.Chem.MolFromSmiles(mol,True)
if not chemMol:
chemMol = rdk.Chem.MolFromSmiles(mol,False)
fingerPrint = rdk.AllChem.GetMorganFingerprint(chemMol,radius)
resDict = fingerPrint.GetNonzeroElements()
except:
continue
fingerPrintsRes[mol] = {}
for ID in resDict:
count = resDict[ID]
name = toolkitsDef["rdk"]["tag"]+"FP_"+str(ID)
if name not in [x.name for x in fingerPrintsAttrs]:
fingerPrintsAttrs.append(orange.FloatVariable(name))
fingerPrintsRes[mol][name] = float(count)
#Add FP attributes even if there was no reference to it. Models will need it as FP not present, i.e. equal 0.0 !
for fpDesc in FP_desc:
name = toolkitsDef["rdk"]["tag"]+fpDesc
if name not in [str(attr.name) for attr in fingerPrintsAttrs]:
fingerPrintsAttrs.append(orange.FloatVariable(name))
#Test attrTypes
for ex in data:
try:
attrObj = []
molStr = str(ex[smilesName].value)
chemMol = rdk.Chem.MolFromSmiles(molStr,True)
if not chemMol:
chemMol = rdk.Chem.MolFromSmiles(molStr,False)
mol = rdk.readstring("mol", rdk.Chem.MolToMolBlock(chemMol))
moldesc = mol.calcdesc(myDescList)
for desc in myDescList:
if type(moldesc[desc]) == str:
attrObj.append(orange.StringVariable(toolkitsDef["rdk"]["tag"] + desc))
else:
attrObj.append(orange.FloatVariable(toolkitsDef["rdk"]["tag"] + desc))
#Process fingerprints
if FingerPrints:
for desc in [fp for fp in fingerPrintsAttrs if fp.name not in attrObj]:
attrObj.append(desc)#orange.FloatVariable(desc.name))
break
except:
continue
resData = orange.ExampleTable(orange.Domain([data.domain[smilesName]] + attrObj,0))
badCompounds = 0
for ex in data:
newEx = orange.Example(resData.domain) # All attrs: ?, ?, ?, ..., ?
newEx[smilesName] = ex[smilesName]
molStr = str(newEx[smilesName].value)
# OBS - add something keeping count on the number of unused smiles
try:
chemMol = rdk.Chem.MolFromSmiles(molStr,True)
if not chemMol:
chemMol = rdk.Chem.MolFromSmiles(molStr,False)
mol = rdk.readstring("mol", rdk.Chem.MolToMolBlock(chemMol))
#mol = rdk.readstring("smi", molStr)
moldesc = mol.calcdesc(myDescList)
for desc in myDescList:
newEx[toolkitsDef["rdk"]["tag"]+desc] = moldesc[desc]
#Process fingerprints
if FingerPrints:
for desc in fingerPrintsAttrs:
if desc.name in fingerPrintsRes[molStr]:
newEx[desc.name] = fingerPrintsRes[molStr][desc.name]
else:
newEx[desc.name] = 0.0
resData.append(newEx)
except:
#print "Unexpected error:", sys.exc_info()
badCompounds += 1
print "Compounds in original data: ",len(data)
print "Compounds able to calculate descs:",len(resData)
print "Ignored Compounds: ",badCompounds
return resData
def getRdkDescResult(data,descList, radius = 1):
""" Calculates the descriptors for the descList using RDK
It expects an attribute containing smiles with a name defined in AZOrangeConfig.SMILESNAMES
It returns a dataset with the same smiles input variable, and as many variables as the descriptors
returned by the toolkit
"""
if "rdk" not in toolkitsEnabled:
return None
FingerPrints = False
smilesName = getSMILESAttr(data)
if not smilesName: return None
FP_desc = []
myDescList = [desc.replace(toolkitsDef["rdk"]["tag"],"") for desc in descList if toolkitsDef["rdk"]["tag"] in desc]
if not myDescList: return None
if "FingerPrints" in myDescList:
FingerPrints = True
myDescList.remove("FingerPrints")
if sum(["FP_" in fp for fp in myDescList]):
tmpDescList = []
FingerPrints = True
for attr in myDescList:
if "FP_" not in attr:
tmpDescList.append(attr)
else:
FP_desc.append(attr)
myDescList = tmpDescList
#Get fingerprints in advance
fingerPrintsAttrs = []
fingerPrintsRes = {}
if FingerPrints:
for ex in data:
mol = str(ex[smilesName].value)
try:
chemMol = rdk.Chem.MolFromSmiles(mol,True)
if not chemMol:
chemMol = rdk.Chem.MolFromSmiles(mol,False)
fingerPrint = rdk.AllChem.GetMorganFingerprint(chemMol,radius)
resDict = fingerPrint.GetNonzeroElements()
except:
continue
fingerPrintsRes[mol] = {}
for ID in resDict:
count = resDict[ID]
name = toolkitsDef["rdk"]["tag"]+"FP_"+str(ID)
if name not in [x.name for x in fingerPrintsAttrs]:
fingerPrintsAttrs.append(orange.FloatVariable(name))
fingerPrintsRes[mol][name] = float(count)
#Add FP attributes even if there was no reference to it. Models will need it as FP not present, i.e. equal 0.0 !
for fpDesc in FP_desc:
name = toolkitsDef["rdk"]["tag"]+fpDesc
if name not in [str(attr.name) for attr in fingerPrintsAttrs]:
fingerPrintsAttrs.append(orange.FloatVariable(name))
#Test attrTypes
for ex in data:
try:
attrObj = []
molStr = str(ex[smilesName].value)
chemMol = rdk.Chem.MolFromSmiles(molStr,True)
if not chemMol:
chemMol = rdk.Chem.MolFromSmiles(molStr,False)
mol = rdk.readstring("mol", rdk.Chem.MolToMolBlock(chemMol))
moldesc = mol.calcdesc(myDescList)
for desc in myDescList:
if type(moldesc[desc]) == str:
attrObj.append(orange.StringVariable(toolkitsDef["rdk"]["tag"] + desc))
else:
attrObj.append(orange.FloatVariable(toolkitsDef["rdk"]["tag"] + desc))
#Process fingerprints
if FingerPrints:
for desc in [fp for fp in fingerPrintsAttrs if fp.name not in attrObj]:
attrObj.append(desc)#orange.FloatVariable(desc.name))
break
except:
continue
resData = orange.ExampleTable(orange.Domain([data.domain[smilesName]] + attrObj,0))
badCompounds = 0
for ex in data:
newEx = orange.Example(resData.domain) # All attrs: ?, ?, ?, ..., ?
newEx[smilesName] = ex[smilesName]
molStr = str(newEx[smilesName].value)
# OBS - add something keeping count on the number of unused smiles
try:
chemMol = rdk.Chem.MolFromSmiles(molStr,True)
if not chemMol:
chemMol = rdk.Chem.MolFromSmiles(molStr,False)
mol = rdk.readstring("mol", rdk.Chem.MolToMolBlock(chemMol))
#mol = rdk.readstring("smi", molStr)
moldesc = mol.calcdesc(myDescList)
for desc in myDescList:
newEx[toolkitsDef["rdk"]["tag"]+desc] = moldesc[desc]
#Process fingerprints
if FingerPrints:
for desc in fingerPrintsAttrs:
if desc.name in fingerPrintsRes[molStr]:
newEx[desc.name] = fingerPrintsRes[molStr][desc.name]
else:
newEx[desc.name] = 0.0
resData.append(newEx)
except:
badCompounds += 1
print "Compounds in original data: ",len(data)
print "Compounds able to calculate descs:",len(resData)
print "Ignored Compounds: ",badCompounds
return resData
