def compute_one_inchi(mol):
molfrom = Chem.MolFromSmiles(mol)
if molfrom == None:
return ((None, None, None))
ini = inchi.MolToInchi(molfrom)
inikey = inchi.InchiToInchiKey(ini)
hsh = hashlib.md5(ini.encode('utf-8')).hexdigest()
return ((hsh, inikey, ini))
def id_cleanup(df, version):
df = df.dropna(subset=["InChIKey"]).reset_index(drop=True)
df.InChIKey = df.InChIKey.apply(lambda x: inchi.InchiToInchiKey(x) if (
x.startswith("InChI")) else x).apply(lambda x: str(x)[:14])
df.broad_id = df.broad_id.apply(lambda x: str(x)[:13])
df.deprecated_broad_id = df.deprecated_broad_id.apply(
lambda x: str(x)[:13])
df = (df.drop_duplicates(
["InChIKey", "pert_iname", "broad_id", "deprecated_broad_id"]).rename(
columns={
"pert_iname": f"pert_iname_{version}",
"broad_id": f"broad_id_{version}",
"deprecated_broad_id": f"deprecated_broad_id_{version}",
"InChIKey": "InChIKey14",
}).reset_index(drop=True))
return df
def parse_f(f):
names = ['']
cid = -1
CAS = f.split('/')[1] if '/' in f else f
CAS = CAS.split('.')[0]
if CAS in ignored_CASs:
return None
failed_mol = False
try:
if CAS in syn_data:
d = syn_data[CAS]
if 'pubchem' in d:
raise Exception(
'Pubchem specified, not trying to use the mol file')
elif 'formula' in d:
raise Exception(
'Formula specified, not trying to use the mol file')
try:
mol = Chem.MolFromMolFile(f)
assert mol is not None
except:
print('Cannot read %s' % f)
1 / 0
try:
inchi_val = inchi.MolToInchi(mol)
except:
print('BAILING ON %s' % f)
1 / 0
mol = inchi.MolFromInchi(inchi_val) # Works better for ions
if mol is None:
print('BAILING ON reconversion to mol %s' % f)
1 / 0
except:
failed_mol = True
if CAS in syn_data:
d = syn_data[CAS]
if 'pubchem' in d:
if str(d['pubchem']) in mycache:
cid, iupac_name, names, mw, smi, inchi_val, inchikey, formula = mycache[
str(d['pubchem'])]
else:
pc = Compound.from_cid(d['pubchem'])
cid = pc.cid
iupac_name = pc.iupac_name
names = pc.synonyms
mw = pc.molecular_weight
smi = pc.canonical_smiles
inchi_val = pc.inchi
inchikey = pc.inchikey
formula = pc.molecular_formula
mycache[str(d['pubchem'])] = (cid, iupac_name, names, mw,
smi, inchi_val, inchikey,
formula)
else:
cid = -1
names = d['synonyms'] if 'synonyms' in d else ['']
mw = float(d['MW'])
smi = d['smiles'] if 'smiles' in d else ''
formula = d['formula'] if 'formula' in d else ''
inchi_val = d['inchi'] if 'inchi' in d else ''
inchikey = d['inchikey'] if 'inchikey' in d else ''
iupac_name = ''
else:
print('FAILED on %s and no custom data was available either' % CAS)
return None
if not failed_mol:
smi = Chem.MolToSmiles(mol, True)
inchi_val = inchi.MolToInchi(mol)
inchikey = inchi.InchiToInchiKey(inchi_val)
mw = Descriptors.MolWt(mol)
# for i in mol.GetAtoms():
# if i.GetIsotope():
# mw = Descriptors.ExactMolWt(mol)
# break
formula = CalcMolFormula(mol, True, True)
iupac_name = ''
try:
if not failed_mol:
if str(inchikey) in mycache:
cid, iupac_name, names = mycache[str(inchikey)]
else:
try:
pc = get_compounds(inchikey, 'inchikey')[0]
cid = pc.cid
iupac_name = pc.iupac_name
names = pc.synonyms
mycache[str(inchikey)] = (cid, iupac_name, names)
except:
mycache[str(inchikey)] = (-1, '', [''])
except:
cid = -1
iupac_name = ''
names = ['']
other_CAS = []
if CAS in pdf_data:
d = pdf_data[CAS]
name = d['Name']
if 'Other Names' in d:
syns = d['Other Names']
else:
syns = []
if not iupac_name:
iupac_name = name
else:
syns.insert(0, name)
if 'Deleted CAS' in d:
other_CAS.extend(d['Deleted CAS'])
if 'Alternate CAS' in d:
other_CAS.extend(d['Alternate CAS'])
syns = [i for i in syns if i not in dup_names]
names = syns + [i for i in names if i not in all_names] + other_CAS
actual_names = []
for name in names:
if name in all_user_names:
# If the name is in the user db, only add it if it corresponds to this CAS number
if CAS in syn_data and 'synonyms' in syn_data[
CAS] and name in syn_data[CAS]['synonyms']:
actual_names.append(name)
else:
# Discard it otherwise
pass
else:
# If the name is not in the user db we're all good
actual_names.append(name)
if CAS in syn_data and 'synonyms' in syn_data[CAS]:
# If the user has any syns for this cas number, add those names if the name hasn't already been aded
for n in syn_data[CAS]['synonyms']:
if n not in actual_names:
actual_names.append(n)
actual_names = [i for i in actual_names if i]
if inchi_val is not None:
inchi_val = inchi_val.replace('InChI=1S/', '')
formula = serialize_formula(formula)
s = '%d\t%s\t%s\t%g\t%s\t%s\t%s\t%s\t' % (cid, CAS, formula, mw, smi,
inchi_val, inchikey, iupac_name)
s += '\t'.join(actual_names)
print(s)
return None
help = []
help.append(counter)
chembl_counter = 0
exister = 0
if (len(bdb_help) != 0) & (len(chembl_help) != 0):
for j in chembl_help:
if bdb_help[0][0] == j[0]:
exister = 1
break
chembl_counter = chembl_counter + 1
#if in chembl
if (exister == 1) & (len(bdb_help) != 0):
help.append(bdb_help[0][0])
test = bdb_help[0][0]
inchi_key = inchi.InchiToInchiKey(bdb_help[0][0])
help.append(inchi_key)
help.append(bdb_help[0][1])
help.append(chembl_help[chembl_counter][1])
mol = inchi.MolFromInchi(bdb_help[0][0], sanitize=False)
smiles = chem.MolToSmiles(mol)
help.append(smiles)
chembl_help.remove(chembl_help[chembl_counter])
bdb_help.remove(bdb_help[0])
#not in chembl, but in bdb
elif (exister == 0) & (len(bdb_help) != 0):
help.append(bdb_help[0][0])
test = bdb_help[0][0]
inchi_key = inchi.InchiToInchiKey(bdb_help[0][0])
help.append(inchi_key)
def _dbize(ec, org, rxn, cof, all_smiles):
"""Place data into MongoDB."""
#Connect to mongodb.
client = MongoClient()
db = client.BrendaDB
ec_collection = db.ec_pages
rxn_collection = db.rxn_pages
cpd_collection = db.cpd_pages
#Build dictionary of reactions and organisms
r_o_dict = {}
for k, v in rxn.iteritems():
p_ = []
r_ = []
#The substrates/products are in name format from the dump, so adding smiles data here.
if len(v[1]) > 0 and len(v[2]) > 0:
for comp in v[1]:
if comp in all_smiles:
smiles = all_smiles[str(comp)]
id = hashlib.sha1(smiles).hexdigest()
inchi = pybel.readstring(
'smi', smiles).write('inchi').strip('\t\n')
inchikey = rdki.InchiToInchiKey(inchi)
r_.append(id)
cpd_collection.update({"_id": id}, {
"$set": {
"smiles": smiles,
"inchi": inchi,
"inchikey": inchikey,
"name": comp
}
},
upsert=True)
else:
r_.append('')
for comp in v[2]:
if comp in all_smiles:
smiles = all_smiles[str(comp)]
id = hashlib.sha1(smiles).hexdigest()
inchi = pybel.readstring(
'smi', smiles).write('inchi').strip('\t\n')
inchikey = rdki.InchiToInchiKey(inchi)
p_.append(id)
cpd_collection.update({"_id": id}, {
"$set": {
"smiles": smiles,
"inchi": inchi,
"inchikey": inchikey,
"name": comp
}
},
upsert=True)
else:
p_.append('')
#A reaction doc is generated containing the names/smiles of both products and reactants as well as a
#stoichiometry vector. The id field is a hash of the final dictionary, and gets added into the rxn/org dict
#for inclusion in the ec pages. Upsert option adds to anything that matches the query and creates a new
#entry if there is no match.
r_entry = {
"r_name": v[1],
"p_name": v[2],
"r_smiles": r_,
"p_smiles": p_,
"s": v[3]
}
rxn_collection.update(
{"_id": hashlib.sha1(str(r_entry)).hexdigest()},
{"$set": {
"rxn": r_entry
}},
upsert=True)
r_o_dict[k] = (v[0], hashlib.sha1(str(r_entry)).hexdigest())
else:
continue
#Iterate through a dictionary of organisms to create the ec pages. Each doc is for a particular organism and lists
#all of the ecs present in it, followed by a list of reactions in each ec listing, with cofactors.
for k, v in org.iteritems():
rxns_in = [x[1] for x in r_o_dict.values() if k in x[0]]
cofs_in = [{"name": x[1], "link": ''} for x in cof if k in x[0]]
for d in cofs_in:
if d["name"] in all_smiles:
d["link"] = hashlib.sha1(all_smiles[str(
d["name"])]).hexdigest()
else:
d["link"] = ''
ec_collection.update({"org": v}, {
"$set": {
"ec." + ec.replace('.', '_'): {
"rxns": rxns_in,
"cofactors": cofs_in
}
}
},
upsert=True)
# ## Merge the Samples and Drugs data
# In[7]:
combined_df = drug_df.merge(sample_df, on="pert_iname", how="inner").reset_index(
drop=True
)
# Move broad_id to first column
col_order = combined_df.columns.tolist()
col_order.insert(0, col_order.pop(col_order.index("broad_id")))
combined_df = combined_df.loc[:, col_order].assign(
InChIKey14=combined_df.InChIKey.apply(
lambda x: inchi.InchiToInchiKey(x) if (x.startswith("InChI")) else x
).apply(lambda x: str(x)[:14])
)
# Output to file
output_file = "repurposing_info"
combined_df.to_csv(f"{output_file}.tsv", sep="\t", index=False)
print(combined_df.shape)
combined_df.head()
# ## Create a "Long" version where we split MOA and Target delimiters
#
# Certain compounds have multiple MOA classes and targets that are delimited by pipes (`|`).
# Each MOA class and target can be considered to have equal support (see https://github.com/broadinstitute/lincs-cell-painting/issues/5).
