def gen_rl_attach(mol='Oc1nc(O)c2nc3c{0}c{1}c{2}c{3}c3nc2n1',
r_l=['', '(S(O)(=O)=O)'],
disp=False,
graph=False):
"""
generate molecules with R group fragment
get_r_list becomes get_multi_r_list so as to generate molecules with multiple R-group attached.
"""
N_group = len(re.findall('{[0-9]*}',
mol)) # find number of R group positions
pdr = get_multi_r_list(N_group, r_l, disp=disp, pdForm=True)
so_l = pdr['Rgroups'].tolist()
aso_l = []
for so in so_l:
aso = mol.format(*so)
aso_l.append(aso)
if disp: print(so, aso)
if graph: jchem.show_mol(aso)
pdr['SMILES'] = aso_l
pdr['BaseMol'] = [aso_l[0]] * len(aso_l)
pdr['BaseStr'] = [mol] * len(aso_l)
return pdr
def gen_rl_2attach( mol, mol_nH, r_l = ['', '(S(O)(=O)=O)'], disp = False, graph = False):
"""
generate molecules with R group fragment
get_r_list becomes get_multi_r_list so as to generate molecules with multiple R-group attached.
Reduced (or hydrated) SMILES strings will be generated as well.
"""
N_group = len( re.findall( '{[0-9]*}', mol)) # find number of R group positions
pdr = get_multi_r_list( N_group, r_l, disp = disp, pdForm = True)
so_l = pdr['Rgroups'].tolist()
aso_l = []
aso_nH_l = []
for so in so_l:
aso = mol.format(*so)
aso_l.append( aso)
aso_nH = mol_nH.format(*so)
aso_nH_l.append( aso_nH)
if disp: print(so, aso, aso_nH)
if graph:
print("Oxidated molecule:")
jchem.show_mol( aso)
print("Hydrated molecule:")
jchem.show_mol( aso_nH)
# Storing canonical smiles strings
pdr['SMILES'] = jchem.csmiles_l( aso_l)
pdr['R-SMILES'] = jchem.csmiles_l( aso_nH_l)
pdr['BaseMol'] = [aso_l[0]] * len( aso_l)
pdr['BaseStr'] = [mol] * len( aso_l)
return pdr
def get_sa( self, s):
m = Chem.MolFromSmiles( s)
sa = sascorer.calculateScore( m)
if self.graph: jchem.show_mol( s)
if self.disp: print('NP Score is', sa)
return sa
def get_np( self, s):
fscore = self.fscore
m = Chem.MolFromSmiles( s)
np = npscorer.scoreMol(m,fscore)
if self.graph: jchem.show_mol( s)
if self.disp: print('NP Score is', np)
return np
def show_Alloxazine():
print('Original Alloxazine')
jchem.show_mol( 'O=C1NC(=O)C2=NC3=CC=CC=C3N=C2N1')
jchem.show_mol( 'O=C1N{0}C(=O)C2=NC3=C{1}C{2}=C{3}C{4}=C3N=C2N1{5}'.format( '(O)','(O)','(O)','(O)','(O)','(O)'))
print('Hydro Alloxazine')
jchem.show_mol( 'O=C1NC2NC3=CC=CC=C3NC2C(=O)N1')
jchem.show_mol( 'O=C1N{5}C2NC3=C{4}C{3}=C{2}C{1}=C3NC2C(=O)N1{0}'.format( '(O)','(O)','(O)','(O)','(O)','(O)'))
def _gen_27aqds_with_oh_r1( self, Frag6_D, show = True):
"""
2,7-AQDS with OH attachment are performed
using smiles interpolation
"""
mol_smiles_list = []
for ix, mol_symb in enumerate(Frag6_D):
mol = bq14_oh2 = Chem.MolFromSmiles( 'C1(O)c2c{B3}c{B4}c(S(=O)(=O)O)c{B5}c2C(O)c2c{B0}c(S(=O)(=O)O)c{B1}c{B2}c21', replacements=mol_symb)
mol_smiles = Chem.MolToSmiles( mol)
mol_smiles_list.append( mol_smiles)
if show:
print(ix+1, mol_smiles)
jchem.show_mol( mol_smiles)
return mol_smiles_list
def show_Alloxazine():
print('Original Alloxazine')
jchem.show_mol('O=C1NC(=O)C2=NC3=CC=CC=C3N=C2N1')
jchem.show_mol('O=C1N{0}C(=O)C2=NC3=C{1}C{2}=C{3}C{4}=C3N=C2N1{5}'.format(
'(O)', '(O)', '(O)', '(O)', '(O)', '(O)'))
print('Hydro Alloxazine')
jchem.show_mol('O=C1NC2NC3=CC=CC=C3NC2C(=O)N1')
jchem.show_mol('O=C1N{5}C2NC3=C{4}C{3}=C{2}C{1}=C3NC2C(=O)N1{0}'.format(
'(O)', '(O)', '(O)', '(O)', '(O)', '(O)'))
def _gen_27aqds_with_oh_r1(self, Frag6_D, show=True):
"""
2,7-AQDS with OH attachment are performed
using smiles interpolation
"""
mol_smiles_list = []
for ix, mol_symb in enumerate(Frag6_D):
mol = bq14_oh2 = Chem.MolFromSmiles(
'C1(O)c2c{B3}c{B4}c(S(=O)(=O)O)c{B5}c2C(O)c2c{B0}c(S(=O)(=O)O)c{B1}c{B2}c21',
replacements=mol_symb)
mol_smiles = Chem.MolToSmiles(mol)
mol_smiles_list.append(mol_smiles)
if show:
print(ix + 1, mol_smiles)
jchem.show_mol(mol_smiles)
return mol_smiles_list
def gen_27aqds_with_R( self, Frag6_D, r_gr, show = True):
"""
2,7-AQDS with OH attachment are performed
using smiles interpolation
"""
mol_smiles_list = []
for ix, mol_symb in enumerate(Frag6_D):
# r_gr = 'S(=O)(=O)O' #[N+]([O-])=O
base_smiles = 'C1(O)c2c{B3}c{B4}c(%s)c{B5}c2C(O)c2c{B0}c(%s)c{B1}c{B2}c21' % (r_gr, r_gr)
mol = Chem.MolFromSmiles( base_smiles, replacements=mol_symb)
mol_smiles = Chem.MolToSmiles( mol)
mol_smiles_list.append( mol_smiles)
if show:
print(ix+1, mol_smiles)
jchem.show_mol( mol_smiles)
return mol_smiles_list
def gen_27aqds_with_R(self, Frag6_D, r_gr, show=True):
"""
2,7-AQDS with OH attachment are performed
using smiles interpolation
"""
mol_smiles_list = []
for ix, mol_symb in enumerate(Frag6_D):
# r_gr = 'S(=O)(=O)O' #[N+]([O-])=O
base_smiles = 'C1(O)c2c{B3}c{B4}c(%s)c{B5}c2C(O)c2c{B0}c(%s)c{B1}c{B2}c21' % (
r_gr, r_gr)
mol = Chem.MolFromSmiles(base_smiles, replacements=mol_symb)
mol_smiles = Chem.MolToSmiles(mol)
mol_smiles_list.append(mol_smiles)
if show:
print(ix + 1, mol_smiles)
jchem.show_mol(mol_smiles)
return mol_smiles_list
def gen_r_attach_Alloxazine_R123457(so3h='(S(O)(=O)=O)',
disp=False,
graph=False):
"""
generate molecules with R group fragment
"""
# n1{R5}c2nc3c{R1}c{R2}c{R3}c{R4}c3nc2c(=O)n{R7}c1=O
#
N_group = 6 #R1234 5 7 -> 0123 4 5
pdr = get_r_list(N_group, so3h, disp=disp, pdForm=True)
so_l = pdr['Rgroups'].tolist()
aso_l = []
mol_l = []
for so in so_l:
if so[4] != '' and so[5] != '':
aso = 'n1{4}c2nc3c{0}c{1}c{2}c{3}c3nc2c(=O)n{5}c1=O'.format(*so)
mol_l.append('n1{4}c2nc3c{0}c{1}c{2}c{3}c3nc2c(=O)n{5}c1=O')
elif so[4] == '' and so[5] == '':
aso = '[nH]1c2nc3c{0}c{1}c{2}c{3}c3nc2c(=O)[nH]c1=O'.format(
*so[:4])
mol_l.append('[nH]1c2nc3c{0}c{1}c{2}c{3}c3nc2c(=O)[nH]c1=O')
elif so[4] == '':
aso = '[nH]1c2nc3c{0}c{1}c{2}c{3}c3nc2c(=O)n{4}c1=O'.format(
so[0], so[1], so[2], so[3], so[5])
mol_l.append('[nH]1c2nc3c{0}c{1}c{2}c{3}c3nc2c(=O)n{4}c1=O')
else: #so[5] == '':
aso = 'n1{4}c2nc3c{0}c{1}c{2}c{3}c3nc2c(=O)[nH]c1=O'.format(
*so[:5])
mol_l.append('n1{4}c2nc3c{0}c{1}c{2}c{3}c3nc2c(=O)[nH]c1=O')
aso_l.append(aso)
if disp: print(so, aso)
if graph: jchem.show_mol(aso)
pdr['SMILES'] = aso_l
pdr['BaseMol'] = [aso_l[0]] * len(aso_l)
pdr['BaseStr'] = mol_l
return pdr
def gen_r_attach( mol = 'Oc1nc(O)c2nc3c{0}c{1}c{2}c{3}c3nc2n1', so3h = '(S(O)(=O)=O)', disp = False, graph = False):
"""
generate molecules with R group fragment
"""
N_group = len( re.findall( '{[0-9]*}', mol)) # find number of R group positions
pdr = get_r_list( N_group, so3h, disp = disp, pdForm = True)
so_l = pdr['Rgroups'].tolist()
aso_l = []
for so in so_l:
aso = mol.format(*so)
aso_l.append( aso)
if disp: print(so, aso)
if graph: jchem.show_mol( aso)
pdr['SMILES'] = aso_l
pdr['BaseMol'] = [aso_l[0]] * len( aso_l)
pdr['BaseStr'] = [mol] * len( aso_l)
return pdr
def bq1x(x='(S(=O)(=O)O)', disp=False):
"""
It generates new quinone molecules with mono functional group attachment.
For anthraquinone(AQ), only two attachment positions are unique and
All the other position attachment generate a duplicated result.
"""
MaxAttach = 1
cs_l = []
en_mol = [''] * MaxAttach
for ix in range(MaxAttach):
en_mol[ix] = x
s = 'C1=CC(=O)C{0}=CC1=O'.format(*en_mol)
cs_l.append(jchem.csmiles(s))
en_mol[ix] = ''
if disp:
print(ix)
jchem.show_mol(s)
return cs_l
def mol1x( mol = 'C1=CC(=O)C{0}=CC1=O', MaxAttach = 1, x = '(S(=O)(=O)O)', disp = False):
"""
It generates new quinone molecules with mono functional group attachment.
For anthraquinone(AQ), only two attachment positions are unique and
All the other position attachment generate a duplicated result.
"""
#MaxAttach = 1
cs_l = []
en_mol = [''] * MaxAttach
for ix in range( MaxAttach):
en_mol[ix] = x
s = mol.format( *en_mol)
cs_l.append( jchem.csmiles( s))
en_mol[ix] = ''
if disp:
print(ix)
jchem.show_mol( s)
return cs_l
def gen_rl_2attach(mol,
mol_nH,
r_l=['', '(S(O)(=O)=O)'],
disp=False,
graph=False):
"""
generate molecules with R group fragment
get_r_list becomes get_multi_r_list so as to generate molecules with multiple R-group attached.
Reduced (or hydrated) SMILES strings will be generated as well.
"""
N_group = len(re.findall('{[0-9]*}',
mol)) # find number of R group positions
pdr = get_multi_r_list(N_group, r_l, disp=disp, pdForm=True)
so_l = pdr['Rgroups'].tolist()
aso_l = []
aso_nH_l = []
for so in so_l:
aso = mol.format(*so)
aso_l.append(aso)
aso_nH = mol_nH.format(*so)
aso_nH_l.append(aso_nH)
if disp: print(so, aso, aso_nH)
if graph:
print("Oxidated molecule:")
jchem.show_mol(aso)
print("Hydrated molecule:")
jchem.show_mol(aso_nH)
# Storing canonical smiles strings
pdr['SMILES'] = jchem.csmiles_l(aso_l)
pdr['R-SMILES'] = jchem.csmiles_l(aso_nH_l)
pdr['BaseMol'] = [aso_l[0]] * len(aso_l)
pdr['BaseStr'] = [mol] * len(aso_l)
return pdr
def aq1x( x = '(S(=O)(=O)O)', disp = False):
"""
It generates new quinone molecules with mono functional group attachment.
For anthraquinone(AQ), only two attachment positions are unique and
All the other position attachment generate a duplicated result.
"""
MaxAttach = 2
cs_l = []
en_mol = [''] * MaxAttach
for ix in range( MaxAttach):
en_mol[ix] = x
s = 'O=C1c2c{0}c{1}ccc2C(=O)c2ccccc21'.format( *en_mol)
cs_l.append( jchem.csmiles( s))
en_mol[ix] = ''
if disp:
print(ix)
jchem.show_mol( s)
return cs_l
def gen_r_attach_Alloxazine_R123457( so3h = '(S(O)(=O)=O)', disp = False, graph = False):
"""
generate molecules with R group fragment
"""
# n1{R5}c2nc3c{R1}c{R2}c{R3}c{R4}c3nc2c(=O)n{R7}c1=O
#
N_group = 6 #R1234 5 7 -> 0123 4 5
pdr = get_r_list( N_group, so3h, disp = disp, pdForm = True)
so_l = pdr['Rgroups'].tolist()
aso_l = []
mol_l = []
for so in so_l:
if so[4] != '' and so[5] != '':
aso = 'n1{4}c2nc3c{0}c{1}c{2}c{3}c3nc2c(=O)n{5}c1=O'.format(*so)
mol_l.append('n1{4}c2nc3c{0}c{1}c{2}c{3}c3nc2c(=O)n{5}c1=O')
elif so[4] == '' and so[5] == '':
aso = '[nH]1c2nc3c{0}c{1}c{2}c{3}c3nc2c(=O)[nH]c1=O'.format(*so[:4])
mol_l.append('[nH]1c2nc3c{0}c{1}c{2}c{3}c3nc2c(=O)[nH]c1=O')
elif so[4] == '':
aso = '[nH]1c2nc3c{0}c{1}c{2}c{3}c3nc2c(=O)n{4}c1=O'.format(so[0],so[1],so[2],so[3], so[5])
mol_l.append('[nH]1c2nc3c{0}c{1}c{2}c{3}c3nc2c(=O)n{4}c1=O')
else: #so[5] == '':
aso = 'n1{4}c2nc3c{0}c{1}c{2}c{3}c3nc2c(=O)[nH]c1=O'.format(*so[:5])
mol_l.append('n1{4}c2nc3c{0}c{1}c{2}c{3}c3nc2c(=O)[nH]c1=O')
aso_l.append( aso)
if disp: print(so, aso)
if graph: jchem.show_mol( aso)
pdr['SMILES'] = aso_l
pdr['BaseMol'] = [aso_l[0]] * len( aso_l)
pdr['BaseStr'] = mol_l
return pdr
def gen_r_attach_lowpot_Flavins(disp=False, graph=False):
oh = '(O)'
h = ''
oc = '(OC)'
rl = []
rl.append(([h, oh, oh, oh, h, h], -0.47))
rl.append(([oh, oh, h, h, h, h], -0.47))
rl.append(([oh, oh, oh, oh, oh, h], -0.47))
rl.append(([oh, oh, oh, oh, h, h], -0.51))
rl.append(([h, oh, h, oh, h, h], -0.50))
rl.append(([h, oh, h, h, h, h], -0.45))
rl.append(([oh, oh, h, oh, oh, h], -0.50))
rl.append(([h, oh, h, oh, oh, h], -0.46))
rl.append(([oh, oh, h, oh, h, h], -0.53))
rl.append(([h, oc, oc, oc, h, h], -0.48))
rl.append(([oc, oc, oc, oc, h, h], -0.48))
rl.append(([oc, oc, h, oc, h, h], -0.47))
rl.append(([h, oc, h, oc, oc, h], -0.46))
rl.append(([oc, oc, h, oc, oc, h], -0.50))
BaseStr = 'n1c2[nH]{5}c3c{4}c{3}c{2}c{1}c3nc2c(=O)[nH]{0}c1=O'
N_group = len(re.findall('{[0-9]*}', BaseStr))
emptyR = [''] * N_group
BaseMol = BaseStr.format(*emptyR)
smiles_l = [BaseStr.format(*r[0]) for r in rl]
pdr = pd.DataFrame()
pdr['ID'] = list(range(1, len(smiles_l) + 1))
R_group_l = []
Index_l = []
NoOfR_l = []
for r in rl:
# Whether it is oh or oc family is determined
r_oh_test = [x == oh for x in r[0]]
print(r[0], '-->', r_oh_test, '-->', any(r_oh_test))
if any(r_oh_test):
r_type = oh
else:
r_type = oc
R_group_l.append(r_type)
r_groups = [0 if x == '' else 1 for x in r[0]]
Index_l.append(r_groups)
NoOfR_l.append(np.sum(r_groups))
pdr['Rgroup'] = R_group_l # This is newly included.
pdr['NoOfR'] = NoOfR_l
pdr['Index'] = Index_l
pdr['Rgroups'] = [r[0] for r in rl]
pdr['SMILES'] = smiles_l
pdr['BaseMol'] = [BaseMol] * len(rl)
pdr['BaseStr'] = [BaseStr] * len(rl)
pdr['RedoxPotential'] = [r[1] for r in rl]
for ix, s in enumerate(smiles_l):
if disp: print(ix + 1, s)
if graph:
jchem.show_mol(s)
return pdr
def gen_r_attach_lowpot_Flavins( disp = False, graph = False):
oh = '(O)'
h = ''
oc = '(OC)'
rl = []
rl.append(([h,oh, oh, oh, h, h], -0.47))
rl.append(([oh, oh, h,h,h,h], -0.47))
rl.append(([oh, oh, oh, oh, oh, h], -0.47))
rl.append(([oh, oh, oh, oh, h, h], -0.51))
rl.append(([h, oh, h, oh, h, h], -0.50))
rl.append(([h, oh, h, h, h, h], -0.45))
rl.append(([oh, oh, h, oh, oh, h], -0.50))
rl.append(([h, oh, h, oh, oh, h], -0.46))
rl.append(([oh, oh, h, oh, h, h], -0.53))
rl.append(([h, oc, oc, oc, h, h], -0.48))
rl.append(([oc, oc, oc, oc, h, h], -0.48))
rl.append(([oc, oc, h, oc, h, h], -0.47))
rl.append(([h, oc, h, oc, oc, h], -0.46))
rl.append(([oc, oc, h, oc, oc, h], -0.50))
BaseStr = 'n1c2[nH]{5}c3c{4}c{3}c{2}c{1}c3nc2c(=O)[nH]{0}c1=O'
N_group = len( re.findall( '{[0-9]*}', BaseStr))
emptyR = [''] * N_group
BaseMol = BaseStr.format( *emptyR)
smiles_l = [ BaseStr.format(*r[0]) for r in rl]
pdr = pd.DataFrame()
pdr['ID'] = list(range( 1, len( smiles_l) + 1))
R_group_l = []
Index_l = []
NoOfR_l = []
for r in rl:
# Whether it is oh or oc family is determined
r_oh_test = [ x == oh for x in r[0]]
print(r[0], '-->', r_oh_test, '-->', any(r_oh_test))
if any(r_oh_test):
r_type = oh
else:
r_type = oc
R_group_l.append( r_type)
r_groups = [ 0 if x == '' else 1 for x in r[0]]
Index_l.append( r_groups)
NoOfR_l.append( np.sum( r_groups))
pdr['Rgroup'] = R_group_l # This is newly included.
pdr['NoOfR'] = NoOfR_l
pdr['Index'] = Index_l
pdr['Rgroups'] = [ r[0] for r in rl]
pdr['SMILES'] = smiles_l
pdr['BaseMol'] = [BaseMol] * len(rl)
pdr['BaseStr'] = [BaseStr] * len(rl)
pdr['RedoxPotential'] = [ r[1] for r in rl]
for ix, s in enumerate( smiles_l):
if disp: print(ix+1, s)
if graph:
jchem.show_mol( s)
return pdr