def test_inorganic_db():
db = ChemicalMetadataDB(elements=False,
main_db=os.path.join(folder, 'Inorganic db.tsv'),
user_dbs=[])
# Check CAS lookup
for CAS, d in db.CAS_index.items():
assert CAS_from_any(d.CASs) == d.CASs
# Try ro check formula lookups
for formula, d in db.formula_index.items():
if formula in set(['H2MgO2', 'F2N2']):
# Formulas which are not unique by design
continue
assert CAS_from_any(formula) == d.CASs
# Check smiles are unique / can lookup by smiles
for smi, d in db.smiles_index.items():
if not smi:
continue
assert CAS_from_any('smiles=' + smi) == d.CASs
# Check formula is formatted right
assert all([i.formula == serialize_formula(i.formula) for i in db.CAS_index.values()])
# Check CAS validity
assert all([checkCAS(i.CASs) for i in db.CAS_index.values()])
# MW checker
for i in db.CAS_index.values():
formula = serialize_formula(i.formula)
atoms = nested_formula_parser(formula, check=False)
mw_calc = molecular_weight(atoms)
assert_allclose(mw_calc, i.MW, atol=0.05)
def test_inorganic_db():
from thermo.identifiers import ChemicalMetadataDB, folder
from thermo.elements import nested_formula_parser, serialize_formula, molecular_weight
db = ChemicalMetadataDB(elements=False,
main_db=os.path.join(folder, 'Inorganic db.tsv'),
user_dbs=[])
for CAS, d in db.CAS_index.items():
assert CAS_from_any(d.CASs) == d.CASs
for formula, d in db.formula_index.items():
if formula in set(['H2MgO2', 'F2N2']):
# Formulas which are not unique by design
continue
assert CAS_from_any(formula) == d.CASs
for smi, d in db.smiles_index.items():
if not smi:
continue
assert CAS_from_any('smiles=' + smi) == d.CASs
assert all([
i.formula == serialize_formula(i.formula)
for i in db.CAS_index.values()
])
assert all([checkCAS(i.CASs) for i in db.CAS_index.values()])
for i in db.CAS_index.values():
formula = serialize_formula(i.formula)
atoms = nested_formula_parser(formula, check=False)
mw_calc = molecular_weight(atoms)
try:
assert_allclose(mw_calc, i.MW, atol=0.05)
except:
print(i)
def test_organic_user_db():
db = ChemicalMetadataDB(elements=False,
main_db=os.path.join(
folder,
'chemical identifiers example user db.tsv'),
user_dbs=[])
for CAS, d in db.CAS_index.items():
assert CAS_from_any(d.CASs) == d.CASs
# Check something was loaded
assert len(db.CAS_index) > 100
# Check smiles are unique / can lookup by smiles
for smi, d in db.smiles_index.items():
if not smi:
continue
assert CAS_from_any('smiles=' + smi) == d.CASs
# Check formula is formatted right
assert all([
i.formula == serialize_formula(i.formula)
for i in db.CAS_index.values()
])
# Check CAS validity
assert all([checkCAS(i.CASs) for i in db.CAS_index.values()])
# MW checker
for i in db.CAS_index.values():
formula = serialize_formula(i.formula)
atoms = nested_formula_parser(formula, check=False)
mw_calc = molecular_weight(atoms)
assert_allclose(mw_calc, i.MW, atol=0.05)
for CAS, d in db.CAS_index.items():
assert CAS_from_any('InChI=1S/' + d.InChI) == int2CAS(CAS)
for CAS, d in db.CAS_index.items():
assert CAS_from_any('InChIKey=' + d.InChI_key) == int2CAS(CAS)
# Test the pubchem ids which aren't -1
for CAS, d in db.CAS_index.items():
if d.pubchemid != -1:
assert CAS_from_any('PubChem=' + str(d.pubchemid)) == int2CAS(CAS)
CAS_lenth = len(db.CAS_index)
assert CAS_lenth == len(db.smiles_index)
assert CAS_lenth == len(db.InChI_index)
assert CAS_lenth == len(db.InChI_key_index)
def test_db_vs_ChemSep():
'''The CAS numbers are checked, as are most of the chemical formulas.
Some chemical structural formulas aren't supported by the current formula
parser and are ignored; otherwise it is a very effective test.
'''
import xml.etree.ElementTree as ET
folder = os.path.join(os.path.dirname(__file__), 'Data')
tree = ET.parse(os.path.join(folder, 'chemsep1.xml'))
root = tree.getroot()
data = {}
for child in root:
CAS = [i.attrib['value'] for i in child if i.tag == 'CAS'][0]
name = [i.attrib['value'] for i in child if i.tag == 'CompoundID'][0]
smiles = [i.attrib['value'] for i in child if i.tag == 'Smiles']
formula = [i.attrib['value'] for i in child if i.tag == 'StructureFormula'][0]
try:
if '-' in formula:
formula = None
else:
formula = serialize_formula(formula)
except:
pass
if smiles:
smiles = smiles[0]
else:
smiles = None
data[CAS] = {'name': name, 'smiles': smiles, 'formula': formula}
for CAS, d in data.items():
hit = pubchem_db.search_CAS(CAS)
assert hit.CASs == CAS
for CAS, d in data.items():
assert CAS_from_any(CAS) == CAS
for CAS, d in data.items():
f = d['formula']
if f is None or f == '1,4-COOH(C6H4)COOH' or d['name'] == 'Air':
continue
assert pubchem_db.search_CAS(CAS).formula == f
# In an ideal world, the names would match too but ~22 don't. Adding more synonyms
# might help.
# Some of them are straight disagreements however
# for CAS, d in data.items():
# try:
# assert CAS_from_any(d['name']) == CAS
# except:
# print(CAS, d['name'])
##
# In an ideal world we could also validate against their smiles
# but that's proving difficult due to things like 1-hexene -
# is it 'CCCCC=C' or 'C=CCCCC'?
#test_db_vs_ChemSep()
def test_organic_user_db():
db = ChemicalMetadataDB(elements=False,
main_db=os.path.join(folder, 'chemical identifiers example user db.tsv'),
user_dbs=[])
for CAS, d in db.CAS_index.items():
assert CAS_from_any(d.CASs) == d.CASs
# Check something was loaded
assert len(db.CAS_index) > 100
# Check smiles are unique / can lookup by smiles
for smi, d in db.smiles_index.items():
if not smi:
continue
assert CAS_from_any('smiles=' + smi) == d.CASs
# Check formula is formatted right
assert all([i.formula == serialize_formula(i.formula) for i in db.CAS_index.values()])
# Check CAS validity
assert all([checkCAS(i.CASs) for i in db.CAS_index.values()])
# MW checker
for i in db.CAS_index.values():
formula = serialize_formula(i.formula)
atoms = nested_formula_parser(formula, check=False)
mw_calc = molecular_weight(atoms)
assert_allclose(mw_calc, i.MW, atol=0.05)
for CAS, d in db.CAS_index.items():
assert CAS_from_any('InChI=1S/' + d.InChI) == int2CAS(CAS)
for CAS, d in db.CAS_index.items():
assert CAS_from_any('InChIKey=' + d.InChI_key) == int2CAS(CAS)
# Test the pubchem ids which aren't -1
for CAS, d in db.CAS_index.items():
if d.pubchemid != -1:
assert CAS_from_any('PubChem=' + str(d.pubchemid)) == int2CAS(CAS)
CAS_lenth = len(db.CAS_index)
assert CAS_lenth == len(db.smiles_index)
assert CAS_lenth == len(db.InChI_index)
assert CAS_lenth == len(db.InChI_key_index)
def test_inorganic_db():
db = ChemicalMetadataDB(elements=False,
main_db=os.path.join(folder, 'Inorganic db.tsv'),
user_dbs=[])
# Check CAS lookup
for CAS, d in db.CAS_index.items():
assert CAS_from_any(d.CASs) == d.CASs
# Try ro check formula lookups
for formula, d in db.formula_index.items():
if formula in set(['H2MgO2', 'F2N2']):
# Formulas which are not unique by design
continue
assert CAS_from_any(formula) == d.CASs
# Check smiles are unique / can lookup by smiles
for smi, d in db.smiles_index.items():
if not smi:
continue
assert CAS_from_any('smiles=' + smi) == d.CASs
# Check formula is formatted right
assert all([
i.formula == serialize_formula(i.formula)
for i in db.CAS_index.values()
])
# Check CAS validity
assert all([checkCAS(i.CASs) for i in db.CAS_index.values()])
# MW checker
for i in db.CAS_index.values():
formula = serialize_formula(i.formula)
atoms = nested_formula_parser(formula, check=False)
mw_calc = molecular_weight(atoms)
assert_allclose(mw_calc, i.MW, atol=0.05)
def test_database_formulas():
# This worked until sisotopes were added to formulas
assert all([
i.formula == serialize_formula(i.formula)
for i in pubchem_db.CAS_index.values()
])
def test_db_vs_ChemSep():
'''The CAS numbers are checked, as are most of the chemical formulas.
Some chemical structural formulas aren't supported by the current formula
parser and are ignored; otherwise it is a very effective test.
'''
import xml.etree.ElementTree as ET
folder = os.path.join(os.path.dirname(__file__), 'Data')
tree = ET.parse(os.path.join(folder, 'chemsep1.xml'))
root = tree.getroot()
data = {}
for child in root:
CAS = [i.attrib['value'] for i in child if i.tag == 'CAS'][0]
name = [i.attrib['value'] for i in child if i.tag == 'CompoundID'][0]
smiles = [i.attrib['value'] for i in child if i.tag == 'Smiles']
formula = [
i.attrib['value'] for i in child if i.tag == 'StructureFormula'
][0]
try:
if '-' in formula:
formula = None
else:
formula = serialize_formula(formula)
except:
pass
if smiles:
smiles = smiles[0]
else:
smiles = None
data[CAS] = {'name': name, 'smiles': smiles, 'formula': formula}
for CAS, d in data.items():
hit = pubchem_db.search_CAS(CAS)
assert hit.CASs == CAS
for CAS, d in data.items():
assert CAS_from_any(CAS) == CAS
for CAS, d in data.items():
f = d['formula']
if f is None or f == '1,4-COOH(C6H4)COOH' or d['name'] == 'Air':
continue
assert pubchem_db.search_CAS(CAS).formula == f
# In an ideal world, the names would match too but ~22 don't. Adding more synonyms
# might help.
# Some of them are straight disagreements however
# for CAS, d in data.items():
# try:
# assert CAS_from_any(d['name']) == CAS
# except:
# print(CAS, d['name'])
##
# In an ideal world we could also validate against their smiles
# but that's proving difficult due to things like 1-hexene -
# is it 'CCCCC=C' or 'C=CCCCC'?
#test_db_vs_ChemSep()
def test_database_formulas():
# This worked until sisotopes were added to formulas
assert all([i.formula == serialize_formula(i.formula) for i in pubchem_db.CAS_index.values()])
def CAS_from_any(ID, autoload=False):
'''Looks up the CAS number of a chemical by searching and testing for the
string being any of the following types of chemical identifiers:
* Name, in IUPAC form or common form or a synonym registered in PubChem
* InChI name, prefixed by 'InChI=1S/' or 'InChI=1/'
* InChI key, prefixed by 'InChIKey='
* PubChem CID, prefixed by 'PubChem='
* SMILES (prefix with 'SMILES=' to ensure smiles parsing; ex.
'C' will return Carbon as it is an element whereas the SMILES
interpretation for 'C' is methane)
* CAS number (obsolete numbers may point to the current number)
If the input is an ID representing an element, the following additional
inputs may be specified as
* Atomic symbol (ex 'Na')
* Atomic number (as a string)
Parameters
----------
ID : str
One of the name formats described above
Returns
-------
CASRN : string
A three-piece, dash-separated set of numbers
Notes
-----
An exception is raised if the name cannot be identified. The PubChem
database includes a wide variety of other synonyms, but these may not be
present for all chemcials.
Examples
--------
>>> CAS_from_any('water')
'7732-18-5'
>>> CAS_from_any('InChI=1S/C2H6O/c1-2-3/h3H,2H2,1H3')
'64-17-5'
>>> CAS_from_any('CCCCCCCCCC')
'124-18-5'
>>> CAS_from_any('InChIKey=LFQSCWFLJHTTHZ-UHFFFAOYSA-N')
'64-17-5'
>>> CAS_from_any('pubchem=702')
'64-17-5'
>>> CAS_from_any('O') # only elements can be specified by symbol
'17778-80-2'
'''
ID = ID.strip()
ID_lower = ID.lower()
if ID in periodic_table:
if periodic_table[ID].number not in homonuclear_elemental_gases:
return periodic_table[ID].CAS
else:
for i in [periodic_table.symbol_to_elements,
periodic_table.number_to_elements,
periodic_table.CAS_to_elements]:
if i == periodic_table.number_to_elements:
if int(ID in i):
return periodic_table[int(ID)].CAS
else:
if ID in i:
return periodic_table[ID].CAS
if checkCAS(ID):
CAS_lookup = pubchem_db.search_CAS(ID, autoload)
if CAS_lookup:
return CAS_lookup.CASs
# handle the case of synonyms
CAS_alternate_loopup = pubchem_db.search_name(ID, autoload)
if CAS_alternate_loopup:
return CAS_alternate_loopup.CASs
if not autoload:
return CAS_from_any(ID, autoload=True)
raise Exception('A valid CAS number was recognized, but is not in the database')
ID_len = len(ID)
if ID_len > 9:
inchi_search = False
# normal upper case is 'InChI=1S/'
if ID_lower[0:9] == 'inchi=1s/':
inchi_search = ID[9:]
elif ID_lower[0:8] == 'inchi=1/':
inchi_search = ID[8:]
if inchi_search:
inchi_lookup = pubchem_db.search_InChI(inchi_search, autoload)
if inchi_lookup:
return inchi_lookup.CASs
else:
if not autoload:
return CAS_from_any(ID, autoload=True)
raise Exception('A valid InChI name was recognized, but it is not in the database')
if ID_lower[0:9] == 'inchikey=':
inchi_key_lookup = pubchem_db.search_InChI_key(ID[9:], autoload)
if inchi_key_lookup:
return inchi_key_lookup.CASs
else:
if not autoload:
return CAS_from_any(ID, autoload=True)
raise Exception('A valid InChI Key was recognized, but it is not in the database')
if ID_len > 8:
if ID_lower[0:8] == 'pubchem=':
pubchem_lookup = pubchem_db.search_pubchem(ID[8:], autoload)
if pubchem_lookup:
return pubchem_lookup.CASs
else:
if not autoload:
return CAS_from_any(ID, autoload=True)
raise Exception('A PubChem integer identifier was recognized, but it is not in the database.')
if ID_len > 7:
if ID_lower[0:7] == 'smiles=':
smiles_lookup = pubchem_db.search_smiles(ID[7:], autoload)
if smiles_lookup:
return smiles_lookup.CASs
else:
if not autoload:
return CAS_from_any(ID, autoload=True)
raise Exception('A SMILES identifier was recognized, but it is not in the database.')
# Try the smiles lookup anyway
# Parsing SMILES is an option, but this is faster
# Pybel API also prints messages to console on failure
smiles_lookup = pubchem_db.search_smiles(ID, autoload)
if smiles_lookup:
return smiles_lookup.CASs
try:
formula_query = pubchem_db.search_formula(serialize_formula(ID), autoload)
if formula_query and type(formula_query) == ChemicalMetadata:
return formula_query.CASs
except:
pass
# Try a direct lookup with the name - the fastest
name_lookup = pubchem_db.search_name(ID, autoload)
if name_lookup:
return name_lookup.CASs
# Permutate through various name options
ID_no_space = ID.replace(' ', '')
ID_no_space_dash = ID_no_space.replace('-', '')
for name in [ID, ID_no_space, ID_no_space_dash]:
for name2 in [name, name.lower()]:
name_lookup = pubchem_db.search_name(name2, autoload)
if name_lookup:
return name_lookup.CASs
if ID[-1] == ')' and '(' in ID:#
# Try to matck in the form 'water (H2O)'
first_identifier, second_identifier = ID[0:-1].split('(', 1)
try:
CAS1 = CAS_from_any(first_identifier)
CAS2 = CAS_from_any(second_identifier)
assert CAS1 == CAS2
return CAS1
except:
pass
if not autoload:
return CAS_from_any(ID, autoload=True)
raise Exception('Chemical name not recognized')
def CAS_from_any(ID, autoload=False):
'''Looks up the CAS number of a chemical by searching and testing for the
string being any of the following types of chemical identifiers:
* Name, in IUPAC form or common form or a synonym registered in PubChem
* InChI name, prefixed by 'InChI=1S/' or 'InChI=1/'
* InChI key, prefixed by 'InChIKey='
* PubChem CID, prefixed by 'PubChem='
* SMILES (prefix with 'SMILES=' to ensure smiles parsing; ex.
'C' will return Carbon as it is an element whereas the SMILES
interpretation for 'C' is methane)
* CAS number (obsolete numbers may point to the current number)
If the input is an ID representing an element, the following additional
inputs may be specified as
* Atomic symbol (ex 'Na')
* Atomic number (as a string)
Parameters
----------
ID : str
One of the name formats described above
Returns
-------
CASRN : string
A three-piece, dash-separated set of numbers
Notes
-----
An exception is raised if the name cannot be identified. The PubChem
database includes a wide variety of other synonyms, but these may not be
present for all chemcials.
Examples
--------
>>> CAS_from_any('water')
'7732-18-5'
>>> CAS_from_any('InChI=1S/C2H6O/c1-2-3/h3H,2H2,1H3')
'64-17-5'
>>> CAS_from_any('CCCCCCCCCC')
'124-18-5'
>>> CAS_from_any('InChIKey=LFQSCWFLJHTTHZ-UHFFFAOYSA-N')
'64-17-5'
>>> CAS_from_any('pubchem=702')
'64-17-5'
>>> CAS_from_any('O') # only elements can be specified by symbol
'17778-80-2'
'''
ID = ID.strip()
ID_lower = ID.lower()
if ID in periodic_table:
if periodic_table[ID].number not in homonuclear_elemental_gases:
return periodic_table[ID].CAS
else:
for i in [periodic_table.symbol_to_elements,
periodic_table.number_to_elements,
periodic_table.CAS_to_elements]:
if i == periodic_table.number_to_elements:
if int(ID in i):
return periodic_table[int(ID)].CAS
else:
if ID in i:
return periodic_table[ID].CAS
if checkCAS(ID):
CAS_lookup = pubchem_db.search_CAS(ID, autoload)
if CAS_lookup:
return CAS_lookup.CASs
# handle the case of synonyms
CAS_alternate_loopup = pubchem_db.search_name(ID, autoload)
if CAS_alternate_loopup:
return CAS_alternate_loopup.CASs
if not autoload:
return CAS_from_any(ID, autoload=True)
raise Exception('A valid CAS number was recognized, but is not in the database')
ID_len = len(ID)
if ID_len > 9:
inchi_search = False
# normal upper case is 'InChI=1S/'
if ID_lower[0:9] == 'inchi=1s/':
inchi_search = ID[9:]
elif ID_lower[0:8] == 'inchi=1/':
inchi_search = ID[8:]
if inchi_search:
inchi_lookup = pubchem_db.search_InChI(inchi_search, autoload)
if inchi_lookup:
return inchi_lookup.CASs
else:
if not autoload:
return CAS_from_any(ID, autoload=True)
raise Exception('A valid InChI name was recognized, but it is not in the database')
if ID_lower[0:9] == 'inchikey=':
inchi_key_lookup = pubchem_db.search_InChI_key(ID[9:], autoload)
if inchi_key_lookup:
return inchi_key_lookup.CASs
else:
if not autoload:
return CAS_from_any(ID, autoload=True)
raise Exception('A valid InChI Key was recognized, but it is not in the database')
if ID_len > 8:
if ID_lower[0:8] == 'pubchem=':
pubchem_lookup = pubchem_db.search_pubchem(ID[8:], autoload)
if pubchem_lookup:
return pubchem_lookup.CASs
else:
if not autoload:
return CAS_from_any(ID, autoload=True)
raise Exception('A PubChem integer identifier was recognized, but it is not in the database.')
if ID_len > 7:
if ID_lower[0:7] == 'smiles=':
smiles_lookup = pubchem_db.search_smiles(ID[7:], autoload)
if smiles_lookup:
return smiles_lookup.CASs
else:
if not autoload:
return CAS_from_any(ID, autoload=True)
raise Exception('A SMILES identifier was recognized, but it is not in the database.')
# Try the smiles lookup anyway
# Parsing SMILES is an option, but this is faster
# Pybel API also prints messages to console on failure
smiles_lookup = pubchem_db.search_smiles(ID, autoload)
if smiles_lookup:
return smiles_lookup.CASs
try:
formula_query = pubchem_db.search_formula(serialize_formula(ID), autoload)
if formula_query and type(formula_query) == ChemicalMetadata:
return formula_query.CASs
except:
pass
# Try a direct lookup with the name - the fastest
name_lookup = pubchem_db.search_name(ID, autoload)
if name_lookup:
return name_lookup.CASs
# Permutate through various name options
ID_no_space = ID.replace(' ', '')
ID_no_space_dash = ID_no_space.replace('-', '')
for name in [ID, ID_no_space, ID_no_space_dash]:
for name2 in [name, name.lower()]:
name_lookup = pubchem_db.search_name(name2, autoload)
if name_lookup:
return name_lookup.CASs
if ID[-1] == ')' and '(' in ID:#
# Try to matck in the form 'water (H2O)'
first_identifier, second_identifier = ID[0:-1].split('(', 1)
try:
CAS1 = CAS_from_any(first_identifier)
CAS2 = CAS_from_any(second_identifier)
assert CAS1 == CAS2
return CAS1
except:
pass
if not autoload:
return CAS_from_any(ID, autoload=True)
raise Exception('Chemical name not recognized')