def _toReactionPNG(rxn):
rc = copy.deepcopy(rxn)
img = Draw.ReactionToImage(rc,
subImgSize=(int(molSize[0] / 3), molSize[1]))
bio = BytesIO()
img.save(bio, format='PNG')
return bio.getvalue()
def _toPNG(mol):
if hasattr(mol, '__sssAtoms'):
highlightAtoms = mol.__sssAtoms
else:
highlightAtoms = []
try:
mol.GetAtomWithIdx(0).GetExplicitValence()
except RuntimeError:
mol.UpdatePropertyCache(False)
if not hasattr(rdMolDraw2D, 'MolDraw2DCairo'):
mc = copy.deepcopy(mol)
try:
img = Draw.MolToImage(mc, size=molSize, kekulize=kekulizeStructures,
highlightAtoms=highlightAtoms)
except ValueError: # <- can happen on a kekulization failure
mc = copy.deepcopy(mol)
img = Draw.MolToImage(mc, size=molSize, kekulize=False, highlightAtoms=highlightAtoms)
bio = BytesIO()
img.save(bio, format='PNG')
return bio.getvalue()
else:
nmol = rdMolDraw2D.PrepareMolForDrawing(mol, kekulize=kekulizeStructures)
d2d = rdMolDraw2D.MolDraw2DCairo(molSize[0], molSize[1])
d2d.DrawMolecule(nmol, highlightAtoms=highlightAtoms)
d2d.FinishDrawing()
return d2d.GetDrawingText()
def _toPNG(mol):
if hasattr(mol, '__sssAtoms'):
highlightAtoms = mol.__sssAtoms
else:
highlightAtoms = []
try:
mol.GetAtomWithIdx(0).GetExplicitValence()
except RuntimeError:
mol.UpdatePropertyCache(False)
if not hasattr(rdMolDraw2D, 'MolDraw2DCairo'):
mc = copy.deepcopy(mol)
try:
img = Draw.MolToImage(mc, size=molSize, kekulize=kekulizeStructures,
highlightAtoms=highlightAtoms)
except ValueError: # <- can happen on a kekulization failure
mc = copy.deepcopy(mol)
img = Draw.MolToImage(mc, size=molSize, kekulize=False, highlightAtoms=highlightAtoms)
bio = BytesIO()
img.save(bio, format='PNG')
return bio.getvalue()
else:
nmol = rdMolDraw2D.PrepareMolForDrawing(mol, kekulize=kekulizeStructures)
d2d = rdMolDraw2D.MolDraw2DCairo(molSize[0], molSize[1])
d2d.DrawMolecule(nmol, highlightAtoms=highlightAtoms)
d2d.FinishDrawing()
return d2d.GetDrawingText()
def _toReactionPNG(rxn):
rc = copy.deepcopy(rxn)
img = Draw.ReactionToImage(rc, subImgSize=(int(molSize[0] / 3), molSize[1]),
highlightByReactant=highlightByReactant)
bio = BytesIO()
img.save(bio, format='PNG')
return bio.getvalue()
def _toPNG(mol):
if hasattr(mol, '__sssAtoms'):
highlightAtoms = mol.__sssAtoms
else:
highlightAtoms = []
try:
mol.GetAtomWithIdx(0).GetExplicitValence()
except RuntimeError:
mol.UpdatePropertyCache(False)
mc = copy.deepcopy(mol)
try:
img = Draw.MolToImage(mc,
size=molSize,
kekulize=kekulizeStructures,
highlightAtoms=highlightAtoms)
except ValueError: # <- can happen on a kekulization failure
mc = copy.deepcopy(mol)
img = Draw.MolToImage(mc,
size=molSize,
kekulize=False,
highlightAtoms=highlightAtoms)
bio = BytesIO()
img.save(bio, format='PNG')
return bio.getvalue()
def getStreamIO(sdfString):
""" Return a StringIO/BytesIO for the string """
if PY3:
sio = BytesIO() if sdfString is None else BytesIO(
sdfString.encode('utf-8'))
else: # pragma: nocover
sio = StringIO() if sdfString is None else StringIO(sdfString)
return sio
def SaveXlsxFromFrame(frame, outFile, molCol='ROMol', size=(300, 300)):
"""
Saves pandas DataFrame as a xlsx file with embedded images.
It maps numpy data types to excel cell types:
int, float -> number
datetime -> datetime
object -> string (limited to 32k character - xlsx limitations)
Cells with compound images are a bit larger than images due to excel.
Column width weirdness explained (from xlsxwriter docs):
The width corresponds to the column width value that is specified in Excel.
It is approximately equal to the length of a string in the default font of Calibri 11.
Unfortunately, there is no way to specify "AutoFit" for a column in the Excel file format.
This feature is only available at runtime from within Excel.
"""
import xlsxwriter # don't want to make this a RDKit dependency
cols = list(frame.columns)
cols.remove(molCol)
dataTypes = dict(frame.dtypes)
workbook = xlsxwriter.Workbook(outFile) # New workbook
worksheet = workbook.add_worksheet() # New work sheet
worksheet.set_column('A:A', size[0] / 6.) # column width
# Write first row with column names
c2 = 1
for x in cols:
worksheet.write_string(0, c2, x)
c2 += 1
c = 1
for _, row in frame.iterrows():
image_data = BytesIO()
img = Draw.MolToImage(row[molCol], size=size)
img.save(image_data, format='PNG')
worksheet.set_row(c, height=size[1]) # looks like height is not in px?
worksheet.insert_image(c, 0, "f", {'image_data': image_data})
c2 = 1
for x in cols:
if str(dataTypes[x]) == "object":
worksheet.write_string(
c, c2,
str(row[x])[:32000]) # string length is limited in xlsx
elif ('float' in str(dataTypes[x])) or ('int' in str(
dataTypes[x])):
if (row[x] != np.nan) or (row[x] != np.inf):
worksheet.write_number(c, c2, row[x])
elif 'datetime' in str(dataTypes[x]):
worksheet.write_datetime(c, c2, row[x])
c2 += 1
c += 1
workbook.close()
image_data.close()
def _get_image(x):
"""displayhook function for PIL Images, rendered as PNG"""
bio = BytesIO()
x.save(bio, format='PNG')
s = b64encode(bio.getvalue()).decode('ascii')
pd.set_option('display.max_columns', len(s) + 1000)
pd.set_option('display.max_rows', len(s) + 1000)
if len(s) + 100 > pd.get_option("display.max_colwidth"):
pd.set_option("display.max_colwidth", len(s) + 1000)
return s
def _get_image(x):
"""displayhook function for PIL Images, rendered as PNG"""
bio = BytesIO()
x.save(bio, format='PNG')
s = b64encode(bio.getvalue()).decode('ascii')
pd.set_option('display.max_columns', len(s) + 1000)
pd.set_option('display.max_rows', len(s) + 1000)
if len(s) + 100 > pd.get_option("display.max_colwidth"):
pd.set_option("display.max_colwidth", len(s) + 1000)
return s
def SaveXlsxFromFrame(frame, outFile, molCol='ROMol', size=(300, 300)):
"""
Saves pandas DataFrame as a xlsx file with embedded images.
It maps numpy data types to excel cell types:
int, float -> number
datetime -> datetime
object -> string (limited to 32k character - xlsx limitations)
Cells with compound images are a bit larger than images due to excel.
Column width weirdness explained (from xlsxwriter docs):
The width corresponds to the column width value that is specified in Excel.
It is approximately equal to the length of a string in the default font of Calibri 11.
Unfortunately, there is no way to specify "AutoFit" for a column in the Excel file format.
This feature is only available at runtime from within Excel.
"""
import xlsxwriter # don't want to make this a RDKit dependency
cols = list(frame.columns)
cols.remove(molCol)
dataTypes = dict(frame.dtypes)
workbook = xlsxwriter.Workbook(outFile) # New workbook
worksheet = workbook.add_worksheet() # New work sheet
worksheet.set_column('A:A', size[0] / 6.) # column width
# Write first row with column names
c2 = 1
for x in cols:
worksheet.write_string(0, c2, x)
c2 += 1
c = 1
for _, row in frame.iterrows():
image_data = BytesIO()
img = Draw.MolToImage(row[molCol], size=size)
img.save(image_data, format='PNG')
worksheet.set_row(c, height=size[1]) # looks like height is not in px?
worksheet.insert_image(c, 0, "f", {'image_data': image_data})
c2 = 1
for x in cols:
if str(dataTypes[x]) == "object":
worksheet.write_string(c, c2, str(row[x])[:32000]) # string length is limited in xlsx
elif ('float' in str(dataTypes[x])) or ('int' in str(dataTypes[x])):
if (row[x] != np.nan) or (row[x] != np.inf):
worksheet.write_number(c, c2, row[x])
elif 'datetime' in str(dataTypes[x]):
worksheet.write_datetime(c, c2, row[x])
c2 += 1
c += 1
workbook.close()
image_data.close()
def _get_image(x):
"""displayhook function for PIL Images, rendered as PNG"""
import pandas as pd
bio = BytesIO()
x.save(bio, format="PNG")
s = b64encode(bio.getvalue()).decode("ascii")
pd.set_option("display.max_columns", len(s) + 1000)
pd.set_option("display.max_rows", len(s) + 1000)
if len(s) + 100 > pd.get_option("display.max_colwidth"):
pd.set_option("display.max_colwidth", len(s) + 1000)
return s
def testSaveState(self):
fName = os.path.join(RDConfig.RDCodeDir, 'ML/Descriptors/test_data', 'molcalc.dsc')
with open(fName, 'r') as inTF:
buf = inTF.read().replace('\r\n', '\n').encode('utf-8')
inTF.close()
inF = BytesIO(buf)
calc = cPickle.load(inF)
self.assertEqual(calc.GetDescriptorNames(), tuple(self.descs))
self.assertEqual(calc.GetDescriptorVersions(), tuple(self.vers))
self._testVals(calc, self.testD)
f = StringIO()
with redirect_stdout(f):
calc.ShowDescriptors()
s = f.getvalue()
for name in calc.GetDescriptorNames():
self.assertIn(name, s)
self.assertIn('Wildman-Crippen LogP value', calc.GetDescriptorSummaries())
self.assertIn('N/A', calc.GetDescriptorSummaries())
funcs = calc.GetDescriptorFuncs()
self.assertEqual(len(funcs), len(self.descs))
for f in funcs:
self.assertTrue(callable(f))
def _loadPackage(self):
with open(os.path.join(self.dataDir, 'Jan9_build3_pkg.pkl'),
'r') as pkgTF:
buf = pkgTF.read().replace('\r\n', '\n').encode('utf-8')
pkgTF.close()
io = BytesIO(buf)
pkg = cPickle.load(io)
return pkg
def _DrawBits(fn, *args, **kwargs):
if 'useSVG' not in kwargs:
kwargs['useSVG'] = ipython_useSVG
res = fn(*args, **kwargs)
if kwargs['useSVG']:
return SVG(res)
else:
sio = BytesIO(res)
return Image.open(sio)
def _toPNG(mol):
if hasattr(mol,'__sssAtoms'):
highlightAtoms=mol.__sssAtoms
else:
highlightAtoms=[]
try:
mol.GetAtomWithIdx(0).GetExplicitValence()
except RuntimeError:
mol.UpdatePropertyCache(False)
mc = copy.deepcopy(mol)
try:
img = Draw.MolToImage(mc,size=molSize,kekulize=kekulizeStructures,
highlightAtoms=highlightAtoms)
except ValueError: # <- can happen on a kekulization failure
mc = copy.deepcopy(mol)
img = Draw.MolToImage(mc,size=molSize,kekulize=False,
highlightAtoms=highlightAtoms)
bio = BytesIO()
img.save(bio,format='PNG')
return bio.getvalue()
def testBuild(self):
# """ tests building and screening a packager """
with open(os.path.join(self.dataDir, 'Jan9_build3_calc.dsc'),
'r') as calcTF:
buf = calcTF.read().replace('\r\n', '\n').encode('utf-8')
calcTF.close()
calc = cPickle.load(BytesIO(buf))
with open(os.path.join(self.dataDir, 'Jan9_build3_model.pkl'),
'rb') as modelF:
model = cPickle.load(modelF)
pkg = Packager.ModelPackage(descCalc=calc, model=model)
self._verify(pkg, self.testD)
def testResults(self):
# " test the results of CrossValidation "
RDRandom.seed(self.randomSeed)
examples, attrs, nPossibleVals = randomtest.GenRandomExamples(
nExamples=200, seed=self.randomArraySeed)
tree, frac = CrossValidate.CrossValidationDriver(examples,
attrs,
nPossibleVals,
silent=1)
self.assertGreater(frac, 0)
with open(self.origTreeName, 'r') as inTFile:
buf = inTFile.read().replace('\r\n', '\n').encode('utf-8')
inTFile.close()
inFile = BytesIO(buf)
oTree = cPickle.load(inFile)
assert oTree == tree, 'Random CrossValidation test failed'
def display_pil_image(img): """displayhook function for PIL Images, rendered as PNG""" bio = BytesIO() img.save(bio, format='PNG') return bio.getvalue()
def display_pil_image(img):
"""displayhook function for PIL Images, rendered as PNG"""
bio = BytesIO()
img.save(bio,format='PNG')
return bio.getvalue()
