def test_cdk(self):
from cinfony import cdk
mol = cdk.readstring("smi", "CCC")
desc = mol.calcdesc(cdk.descs[1:3])
for d in desc:
if desc[d] != desc[d]:
desc[d] = '?'
expectedDesc = {'ip': 0.0, 'weight': 44.062600320000001}
self.assertEqual(desc,expectedDesc)
def hide_test_cdk(self):
from cinfony import cdk
mol = cdk.readstring("smi", "CCC")
desc = mol.calcdesc(cdk.descs[1:3])
for d in desc:
if desc[d] != desc[d]:
desc[d] = '?'
expectedDesc = {'ip': 0.0, 'weight': 44.062600320000001}
self.assertEqual(desc,expectedDesc)
def getCdkDescResult(data,descList):
""" Calculates the descriptors for the descList using cdk
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 "cdk" not in toolkitsEnabled:
return None
smilesName = getSMILESAttr(data)
if not smilesName: return None
myDescList = [desc.replace(cdkTag,"") for desc in descList if cdkTag in desc]
if not myDescList: return None
#Compute the results
results = {}
for ex in data:
smile = str(ex[smilesName].value)
mol = cdk.readstring("smi", smile)
results[smile] = mol.calcdesc(myDescList)
# Get all the different descriptor names returned by cdk
varNames = []
for res in results:
for desc in results[res]:
if desc not in varNames:
varNames.append(desc)
# Generate the dataset assuring the same order of examples
resData = orange.ExampleTable(orange.Domain([data.domain[smilesName]] + [orange.FloatVariable(cdkTag+name) for name in varNames],0))
for ex in data:
newEx = orange.Example(resData.domain)
smile = str(ex[smilesName].value)
newEx[smilesName] =smile
for desc in results[smile]:
newEx[cdkTag+desc] = results[smile][desc]
resData.append(newEx)
return resData
def getCdkDescResult(data,descList):
""" Calculates the descriptors for the descList using cdk
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 "cdk" not in toolkitsEnabled:
return None
smilesName = getSMILESAttr(data)
if not smilesName: return None
myDescList = [desc.replace(toolkitsDef["cdk"]["tag"],"") for desc in descList if toolkitsDef["cdk"]["tag"] in desc]
if not myDescList: return None
#Compute the results
results = {}
for ex in data:
smile = str(ex[smilesName].value)
mol = cdk.readstring("smi", smile)
results[smile] = mol.calcdesc(myDescList)
# Get all the different descriptor names returned by cdk
varNames = []
for res in results:
for desc in results[res]:
if desc not in varNames:
varNames.append(desc)
# Generate the dataset assuring the same order of examples
resData = orange.ExampleTable(orange.Domain([data.domain[smilesName]] + [orange.FloatVariable(toolkitsDef["cdk"]["tag"]+name) for name in varNames],0))
for ex in data:
newEx = orange.Example(resData.domain)
smile = str(ex[smilesName].value)
newEx[smilesName] =smile
for desc in results[smile]:
newEx[toolkitsDef["cdk"]["tag"]+desc] = results[smile][desc]
resData.append(newEx)
return resData
