def testAppend(self):
try:
fname = tempfile.mktemp() + ".smi"
fname2 = tempfile.mktemp() + "-2.smi"
storefname = tempfile.mktemp() + ".store"
with open(fname, 'w') as f:
f.write(one_smiles)
opts = make_store.MakeStorageOptions(storage=storefname,
smilesfile=fname,
hasHeader=False,
batchsize=1,
smilesColumn=0,
nameColumn=1,
seperator=" ",
descriptors="RDKit2DSubset",
index_inchikey=True)
make_store.make_store(opts)
with contextlib.closing(DescriptaStore(storefname)) as store:
self.assertEqual(store.lookupName("0"), 0)
self.assertEqual(
store.lookupInchiKey("UHOVQNZJYSORNB-UHFFFAOYSA-N"), [0])
self.assertEqual(store.descriptors().get(0),
(True, 78.046950192, 0.0, 1.0, 0.0, 1.0))
# now append some junk
with open(fname2, 'w') as f:
f.write(two_smiles)
opts.smilesfile = fname2
append_store.append_store(opts)
with contextlib.closing(DescriptaStore(storefname)) as store:
self.assertEqual(len(store), 2)
self.assertEqual(store.lookupName("0"), 0)
self.assertEqual(
store.lookupInchiKey("UHOVQNZJYSORNB-UHFFFAOYSA-N"),
[0, 1])
self.assertEqual(store.descriptors().get(0),
(True, 78.046950192, 0.0, 1.0, 0.0, 1.0))
self.assertEqual(store.lookupName("1"), 1)
self.assertEqual(
store.lookupInchiKey("UHOVQNZJYSORNB-UHFFFAOYSA-N"),
[0, 1])
self.assertEqual(store.descriptors().get(1),
(True, 78.046950192, 0.0, 1.0, 0.0, 1.0))
finally:
if os.path.exists(fname):
os.unlink(fname)
if os.path.exists(fname2):
os.unlink(fname2)
if os.path.exists(storefname):
shutil.rmtree(storefname)
def testCanonicalSmiles2(self):
try:
fname = tempfile.mktemp() + ".smi"
storefname = tempfile.mktemp() + ".store"
with open(fname, 'w') as f:
f.write(many_smiles)
opts = make_store.MakeStorageOptions(storage=storefname,
smilesfile=fname,
hasHeader=False,
smilesColumn=0,
nameColumn=1,
seperator=" ",
descriptors="Canonicalize",
index_inchikey=False)
make_store.make_store(opts)
with contextlib.closing(DescriptaStore(storefname)) as store:
counts = []
for i in range(10):
r = store.descriptors().get(i)
counts.append(r[0])
counts.sort()
self.assertEqual(counts, list(range(8, 18)))
finally:
if os.path.exists(fname):
os.unlink(fname)
if os.path.exists(storefname):
shutil.rmtree(storefname)
def testChiralMorgans(self):
try:
fname = tempfile.mktemp()+".smi"
storefname = tempfile.mktemp()+".store"
with open(fname, 'w') as f:
f.write(many_smiles),
opts = make_store.MakeStorageOptions( storage=storefname, smilesfile=fname,
hasHeader=False,
smilesColumn=0, nameColumn=1,
seperator=" ", descriptors="MorganChiral3Counts",
index_inchikey=True )
make_store.make_store(opts)
generator = DescriptorGenerator.REGISTRY["MorganChiral3Counts".lower()]
with contextlib.closing(DescriptaStore(storefname)) as store:
for i in range(10):
r = store.descriptors().get(i)
self.assertEqual(r, expected_chiral_data[i])
finally:
if os.path.exists(fname):
os.unlink(fname)
if os.path.exists(storefname):
shutil.rmtree(storefname)
def testNonesWithCalcFlags(self):
try:
fname = tempfile.mktemp() + ".smi"
storefname = tempfile.mktemp() + ".store"
with open(fname, 'w') as f:
f.write(one_smiles)
opts = make_store.MakeStorageOptions(
storage=storefname,
smilesfile=fname,
hasHeader=False,
smilesColumn=0,
nameColumn=1,
seperator=" ",
descriptors="NANDescriptorsWithCalcFlags",
index_inchikey=True)
make_store.make_store(opts)
with contextlib.closing(DescriptaStore(storefname)) as store:
self.assertFalse(store.descriptors().get(0)[0])
finally:
if os.path.exists(fname):
os.unlink(fname)
if os.path.exists(storefname):
shutil.rmtree(storefname)
def testRDKitFPBits(self):
try:
fname = tempfile.mktemp() + ".smi"
storefname = tempfile.mktemp() + ".store"
with open(fname, 'w') as f:
f.write("\n".join(
['{0} {1}'.format(v, k) for k, v in testSmiles.items()]))
opts = make_store.MakeStorageOptions(storage=storefname,
smilesfile=fname,
hasHeader=False,
smilesColumn=0,
nameColumn=1,
seperator=" ",
descriptors="RDKitFPBits",
index_inchikey=False)
make_store.make_store(opts)
with contextlib.closing(DescriptaStore(storefname)) as store:
for i in range(4):
r = store.descriptors().get(i)
self.assertEqual(r, expected_RDKFP_data[i])
finally:
if os.path.exists(fname):
os.unlink(fname)
if os.path.exists(storefname):
shutil.rmtree(storefname)
def testNormalized(self):
try:
fname = tempfile.mktemp() + ".smi"
storefname = tempfile.mktemp() + ".store"
with open(fname, 'w') as f:
f.write(many_smiles),
opts = make_store.MakeStorageOptions(
storage=storefname,
smilesfile=fname,
hasHeader=False,
smilesColumn=0,
nameColumn=1,
seperator=" ",
descriptors="RDKit2DNormalized",
index_inchikey=True)
make_store.make_store(opts)
generator = DescriptorGenerator.REGISTRY[
"RDKit2DNormalized".lower()]
results = []
with contextlib.closing(DescriptaStore(storefname)) as store:
for i in range(10):
r = store.descriptors().get(i)
compare_results(self, r, expected[i])
finally:
if os.path.exists(fname):
os.unlink(fname)
if os.path.exists(storefname):
shutil.rmtree(storefname)
def testManyNoInchi(self):
try:
fname = tempfile.mktemp() + ".smi"
storefname = tempfile.mktemp() + ".store"
with open(fname, 'w') as f:
f.write(many_smiles)
opts = make_store.MakeStorageOptions(storage=storefname,
smilesfile=fname,
hasHeader=False,
batchsize=1,
smilesColumn=0,
nameColumn=1,
seperator=" ",
descriptors="RDKit2DSubset",
index_inchikey=False)
make_store.make_store(opts)
origdata = many_smiles.split("\n")
with contextlib.closing(DescriptaStore(storefname)) as store:
for i in range(10):
self.assertEqual(store.lookupName(str(i)), i)
self.assertEqual(store.descriptors().get(0),
(True, 78.046950192, 0.0, 1.0, 0.0, 1.0))
self.assertEqual(store.descriptors().get(1),
(True, 92.062600256, 0.0, 1.0, 0.0, 1.0))
self.assertEqual(store.descriptors().get(2),
(True, 106.07825032, 0.0, 1.0, 0.0, 1.0))
self.assertEqual(store.descriptors().get(3),
(True, 120.093900384, 0.0, 1.0, 0.0, 1.0))
self.assertEqual(store.descriptors().get(4),
(True, 134.109550448, 0.0, 1.0, 0.0, 1.0))
self.assertEqual(store.descriptors().get(5),
(True, 148.125200512, 0.0, 1.0, 0.0, 1.0))
self.assertEqual(store.descriptors().get(6),
(True, 162.140850576, 0.0, 1.0, 0.0, 1.0))
self.assertEqual(store.descriptors().get(7),
(True, 176.15650064, 0.0, 1.0, 0.0, 1.0))
self.assertEqual(store.descriptors().get(8),
(True, 190.172150704, 0.0, 1.0, 0.0, 1.0))
self.assertEqual(store.descriptors().get(9),
(True, 204.187800768, 0.0, 1.0, 0.0, 1.0))
self.assertEqual(store.descriptors().get(10),
(False, 0.0, 0.0, 0.0, 0.0, 0.0))
for i in range(10):
m = store.molIndex().getRDMol(i)
smiles, name = store.molIndex().get(i)
self.assertEqual(name, str(i))
self.assertEqual(smiles, origdata[i].split()[0])
finally:
if os.path.exists(fname):
os.unlink(fname)
if os.path.exists(storefname):
shutil.rmtree(storefname)
def testColCache(self):
try:
fname = tempfile.mktemp() + ".smi"
storefname = tempfile.mktemp() + ".store"
with open(fname, 'w') as f:
f.write(many_smiles)
opts = make_store.MakeStorageOptions(storage=storefname,
smilesfile=fname,
hasHeader=False,
smilesColumn=0,
nameColumn=1,
seperator=" ",
descriptors="RDKit2DSubset",
index_inchikey=True)
make_store.make_store(opts)
with contextlib.closing(DescriptaStore(storefname)) as store:
cols = []
# get normal data
for idx, _ in enumerate(store.db.colnames):
col = list(store.db.getColByIdx(idx))
cols.append(col)
# cache the columns
store.db.cacheColumns()
# make sure the datafiles are written
for idx, _ in enumerate(store.db.colnames):
fn = os.path.join(store.db.colCacheDir, str(idx))
self.assertTrue(os.path.exists(fn), fn)
for idx, _ in enumerate(store.db.colnames):
col = list(store.db.getColByIdx(idx))
self.assertEqual(col, cols[idx])
# swap a data file
idx = 0, 1
fn0 = os.path.join(store.db.colCacheDir, str(0))
fn1 = os.path.join(store.db.colCacheDir, str(1))
shutil.move(fn0, fn0 + ".bak")
shutil.move(fn1, fn0)
shutil.move(fn0 + ".bak", fn1)
try:
col = list(store.db.getColByIdx(0))
caught = false
except struct.error:
caught = True
self.assertTrue(
caught, "moving cache file should have broken the cache")
finally:
if os.path.exists(fname):
os.unlink(fname)
if os.path.exists(storefname):
shutil.rmtree(storefname)
def main():
opts = parser.parse_args()
store = DescriptaStore(opts.storage)
N = len(store)
gen = store.getDescriptorCalculator()
randomize = True
if opts.samples == -1:
randomize = False
opts.samples = len(store)
next = .05
for i in range(opts.samples):
if i and float(i) / opts.samples > next:
print("Validated %2.2f%%" % (next * 100))
next += .05
if randomize:
idx = random.randint(0, N - 1)
else:
idx = i
v = store.descriptors().get(idx)
smiles = store.molIndex().getMol(idx)
name = None
try:
name = store.molIndex().getName(idx)
except:
pass
res = gen.process(smiles)
if res is None:
assert v == tuple([0] * len(v))
continue
data = []
for x in gen.process(smiles):
if math.isnan(x): data.append(None)
else: data.append(x)
v2 = []
for x in v:
if math.isnan(x): v2.append(None)
else: v2.append(x)
assert v2 == data, "idx:%s smiles:%s name:%s \n%r\n\t%r" % (
idx, smiles, name, v, data)
def testOffByOne(self):
try:
fname = tempfile.mktemp() + ".smi"
storefname = tempfile.mktemp() + ".store"
with open(fname, 'w') as f:
f.write(one_smiles)
opts = make_store.MakeStorageOptions(storage=storefname,
smilesfile=fname,
hasHeader=False,
batchsize=1,
smilesColumn=0,
nameColumn=1,
seperator=" ",
descriptors="RDKit2DSubset",
index_inchikey=True)
make_store.make_store(opts)
with contextlib.closing(DescriptaStore(storefname)) as store:
self.assertEqual(store.lookupName("0"), 0)
self.assertEqual(
store.lookupInchiKey("UHOVQNZJYSORNB-UHFFFAOYSA-N"), [0])
self.assertEqual(store.descriptors().get(0),
(True, 78.046950192, 0.0, 1.0, 0.0, 1.0))
try:
store.lookupInchiKey("MY DOG HAS FLEAS")
self.assertTrue(False) # should not get here
except KeyError:
pass
finally:
if os.path.exists(fname):
os.unlink(fname)
if os.path.exists(storefname):
shutil.rmtree(storefname)
parser.add_argument("--output-name", action="store_true", default=True,
help="outputs the molecule name (if available)")
# to do?
#parser.add_argument("--output-smiles", action="store_true",
# help="outputs the smiles column")
#parser.add_argument("--output-inchi", action="store_true",
# help="outputs an inchi column (if available)")
parser.add_argument("--verbose", action="store_true",
help="Verbose logging")
opts = parser.parse_args()
store = DescriptaStore(opts.storage)
if opts.seperator == ",":
writer = csv.writer(sys.stdout)
elif opts.seperator == "\t":
writer = csv.writer(sys.stdout, dialiect=csv.excel_tab)
if opts.keep_missing_descriptors:
opts.keep_calculatedflags = True
if opts.output_name:
writer.writerow(['Name'] + store.getDescriptorNames(opts.keep_calculatedflags))
else:
writer.writerow(store.getDescriptorNames(opts.keep_calculatedflags))
if not opts.namefile:
indices = range(len(store))
def testAppend(self):
try:
fname = tempfile.mktemp() + ".smi"
storefname = tempfile.mktemp() + ".store"
with open(fname, 'w') as f:
f.write(many_smiles)
opts = make_store.MakeStorageOptions(storage=storefname,
smilesfile=fname,
hasHeader=False,
smilesColumn=0,
nameColumn=1,
seperator=" ",
descriptors="RDKit2DSubset",
index_inchikey=True)
make_store.make_store(opts)
with contextlib.closing(DescriptaStore(storefname)) as store:
for i in range(10):
self.assertEqual(store.lookupName(str(i)), i)
for i in range(10):
m = store.molIndex().getRDMol(i)
inchi = AllChem.InchiToInchiKey(AllChem.MolToInchi(m))
self.assertEqual(store.lookupInchiKey(inchi), [i])
self.assertEqual(store.descriptors().get(0),
(True, 78.046950192, 0.0, 1.0, 0.0, 1.0))
self.assertEqual(store.descriptors().get(1),
(True, 92.062600256, 0.0, 1.0, 0.0, 1.0))
self.assertEqual(store.descriptors().get(2),
(True, 106.07825032, 0.0, 1.0, 0.0, 1.0))
self.assertEqual(store.descriptors().get(3),
(True, 120.093900384, 0.0, 1.0, 0.0, 1.0))
self.assertEqual(store.descriptors().get(4),
(True, 134.109550448, 0.0, 1.0, 0.0, 1.0))
self.assertEqual(store.descriptors().get(5),
(True, 148.125200512, 0.0, 1.0, 0.0, 1.0))
self.assertEqual(store.descriptors().get(6),
(True, 162.140850576, 0.0, 1.0, 0.0, 1.0))
self.assertEqual(store.descriptors().get(7),
(True, 176.15650064, 0.0, 1.0, 0.0, 1.0))
self.assertEqual(store.descriptors().get(8),
(True, 190.172150704, 0.0, 1.0, 0.0, 1.0))
self.assertEqual(store.descriptors().get(9),
(True, 204.187800768, 0.0, 1.0, 0.0, 1.0))
self.assertEqual(store.descriptors().get(10),
(False, 0.0, 0.0, 0.0, 0.0, 0.0))
fname = tempfile.mktemp() + ".smi"
with open(fname, 'w') as f:
f.write(many_smiles2)
opts.smilesfile = fname
append_store.append_store(opts)
with contextlib.closing(DescriptaStore(storefname)) as store:
for i in range(10):
self.assertEqual(store.lookupName(str(i)), i)
for i in range(10):
m = store.molIndex().getRDMol(i)
inchi = AllChem.InchiToInchiKey(AllChem.MolToInchi(m))
m = store.molIndex().getRDMol(i + 11)
self.assertTrue(m != None)
inchi2 = AllChem.InchiToInchiKey(AllChem.MolToInchi(m))
self.assertEqual(inchi, inchi2)
self.assertEqual(store.lookupInchiKey(inchi), [i, i + 11])
for i in range(2):
self.assertEqual(store.descriptors().get(11 + 0),
(True, 78.046950192, 0.0, 1.0, 0.0, 1.0))
self.assertEqual(store.descriptors().get(11 + 1),
(True, 92.062600256, 0.0, 1.0, 0.0, 1.0))
self.assertEqual(store.descriptors().get(11 + 2),
(True, 106.07825032, 0.0, 1.0, 0.0, 1.0))
self.assertEqual(store.descriptors().get(11 + 3),
(True, 120.093900384, 0.0, 1.0, 0.0, 1.0))
self.assertEqual(store.descriptors().get(11 + 4),
(True, 134.109550448, 0.0, 1.0, 0.0, 1.0))
self.assertEqual(store.descriptors().get(11 + 5),
(True, 148.125200512, 0.0, 1.0, 0.0, 1.0))
self.assertEqual(store.descriptors().get(11 + 6),
(True, 162.140850576, 0.0, 1.0, 0.0, 1.0))
self.assertEqual(store.descriptors().get(11 + 7),
(True, 176.15650064, 0.0, 1.0, 0.0, 1.0))
self.assertEqual(store.descriptors().get(11 + 8),
(True, 190.172150704, 0.0, 1.0, 0.0, 1.0))
self.assertEqual(store.descriptors().get(11 + 9),
(True, 204.187800768, 0.0, 1.0, 0.0, 1.0))
self.assertEqual(store.descriptors().get(11 + 10),
(False, 0.0, 0.0, 0.0, 0.0, 0.0))
finally:
if os.path.exists(fname):
os.unlink(fname)
if os.path.exists(storefname):
shutil.rmtree(storefname)
def testMany(self):
try:
fname = tempfile.mktemp() + ".smi"
storefname = tempfile.mktemp() + ".store"
with open(fname, 'w') as f:
f.write(many_smiles)
opts = make_store.MakeStorageOptions(storage=storefname,
smilesfile=fname,
hasHeader=False,
smilesColumn=0,
nameColumn=1,
seperator=" ",
descriptors="RDKit2DSubset",
index_inchikey=True)
make_store.make_store(opts)
with contextlib.closing(DescriptaStore(storefname)) as store:
for i in range(10):
self.assertEqual(store.lookupName(str(i)), i)
self.assertEqual(store.descriptors().get(0),
(True, 78.046950192, 0.0, 1.0, 0.0, 1.0))
self.assertEqual(store.descriptors().get(1),
(True, 92.062600256, 0.0, 1.0, 0.0, 1.0))
self.assertEqual(store.descriptors().get(2),
(True, 106.07825032, 0.0, 1.0, 0.0, 1.0))
self.assertEqual(store.descriptors().get(3),
(True, 120.093900384, 0.0, 1.0, 0.0, 1.0))
self.assertEqual(store.descriptors().get(4),
(True, 134.109550448, 0.0, 1.0, 0.0, 1.0))
self.assertEqual(store.descriptors().get(5),
(True, 148.125200512, 0.0, 1.0, 0.0, 1.0))
self.assertEqual(store.descriptors().get(6),
(True, 162.140850576, 0.0, 1.0, 0.0, 1.0))
self.assertEqual(store.descriptors().get(7),
(True, 176.15650064, 0.0, 1.0, 0.0, 1.0))
self.assertEqual(store.descriptors().get(8),
(True, 190.172150704, 0.0, 1.0, 0.0, 1.0))
self.assertEqual(store.descriptors().get(9),
(True, 204.187800768, 0.0, 1.0, 0.0, 1.0))
self.assertEqual(store.descriptors().get(10),
(False, 0.0, 0.0, 0.0, 0.0, 0.0))
self.assertEqual(
store.descriptors().getDict(7),
toDict((True, 176.15650064, 0.0, 1.0, 0.0, 1.0)))
calc = store.getDescriptorCalculator()
for i in range(10):
m = store.molIndex().getRDMol(i)
sm = AllChem.MolToSmiles(m)
inchi = AllChem.InchiToInchiKey(AllChem.MolToInchi(m))
self.assertEqual(store.lookupInchiKey(inchi), [i])
v = store.descriptors().get(i)
sv = tuple(calc.process(sm))
self.assertEqual(v, sv)
finally:
if os.path.exists(fname):
os.unlink(fname)
if os.path.exists(storefname):
shutil.rmtree(storefname)
from descriptastorus import DescriptaStore
import argparse, os, logging, shutil, time, random
import sys
from rdkit import rdBase
rdBase.DisableLog("rdApp.*")
parser = argparse.ArgumentParser()
parser.add_argument("storage",
help="directory in which to store the descriptors")
parser.add_argument("storage2",
help="directory in which to store the descriptors")
opts = parser.parse_args()
store1 = DescriptaStore(opts.storage)
store2 = DescriptaStore(opts.storage2)
if store1.getDescriptorNames() != store2.getDescriptorNames():
logging.warning("Storages are not compatible, columns are different")
s1 = set(store1.getDescriptorNames())
s2 = set(store2.getDescriptorNames())
if s1 == s2:
logging.warning("Columns are the same but reordered")
else:
logging.warning("Extra columns in store1:\n\t%s", "\n\t".join(s1-s2))
logging.warning("Extra columns in store2:\n\t%s", "\n\t".join(s2-s1))
for i,(a,b) in enumerate(zip(store1.getDescriptorNames(),
store2.getDescriptorNames())):
