def _mcs(data,
asSmiles,
atomCompare,
bondCompare,
threshold,
ringMatchesRingOnly,
completeRingsOnly,
sanitize=True,
removeHs=True,
strictParsing=True,
isomericSmiles=False,
canonical=True,
kekuleSmiles=False):
ms = _parseMolData(data,
sanitize=sanitize,
removeHs=removeHs,
strictParsing=strictParsing)
if not ms:
return
if len(ms) == 1:
if asSmiles:
print 'SMARTS'
return Chem.MolToSmiles(ms[0])
else:
print 'SMILES'
return Chem.MolToSmarts(ms[0])
if threshold:
threshold = float(threshold)
try:
mcs = MCS.FindMCS(ms,
atomCompare=atomCompare,
bondCompare=bondCompare,
ringMatchesRingOnly=ringMatchesRingOnly,
completeRingsOnly=completeRingsOnly,
threshold=threshold)
except TypeError:
mcs = MCS.FindMCS(ms,
atomCompare=atomCompare,
bondCompare=bondCompare,
ringMatchesRingOnly=ringMatchesRingOnly,
completeRingsOnly=completeRingsOnly)
res = mcs.smarts
if asSmiles:
p = Chem.MolFromSmarts(res)
for m in ms:
if m.HasSubstructMatch(p):
match = m.GetSubstructMatch(p)
res = Chem.MolFragmentToSmiles(m,
atomsToUse=match,
isomericSmiles=isomericSmiles,
canonical=canonical,
kekuleSmiles=kekuleSmiles)
break
return res
#-----------------------------------------------------------------------------------------------------------------------
def mcs(self, fragments):
"""Find the maximum common substructure from a list of fragments.
N.B.: Currently does not expose the many options provided by rdkit:
http://www.rdkit.org/Python_Docs/rdkit.Chem.MCS-module.html
Also, SMARTS match naturally includes heavy atoms only.
:param fragments: two or more fragments containing common substructure
:type fragments : list
:return: maximum common substructure result
:rtype : MCSResult
"""
try:
global MCS
global rdk
MCS
except NameError:
from rdkit.Chem import MCS
from cinfony import rdk
rf = [rdk.Molecule(f.molecule).Mol for f in fragments]
cs = MCS.FindMCS(rf)
return cs
def test_timeout_negative(self):
try:
MCS.FindMCS(lengthy_mols, timeout=-1)
except ValueError:
pass
else:
raise AssertionError("bad range check for timeout")
def test_min_atoms_1(self):
try:
result = MCS.FindMCS(simple_mols, minNumAtoms=1)
except ValueError:
pass
else:
raise AssertionError("should have raised an exception")
def _mcs(data, params):
ms = _parseMolData(data)
if not ms:
return
if len(ms) == 1:
if bool(int(params.get('asSmiles', '0'))):
print 'SMARTS'
return Chem.MolToSmiles(ms[0])
else:
print 'SMILES'
return Chem.MolToSmarts(ms[0])
atomCompare = params.get('atomCompare', 'elements')
bondCompare = params.get('bondCompare', 'bondtypes')
ringMatchesRingOnly = bool(int(params.get('ringMatchesRingOnly', '0')))
completeRingsOnly = bool(int(params.get('completeRingsOnly', '0')))
threshold = params.get('threshold', None)
if threshold:
threshold = float(threshold)
try:
mcs = MCS.FindMCS(ms,
atomCompare=atomCompare,
bondCompare=bondCompare,
ringMatchesRingOnly=ringMatchesRingOnly,
completeRingsOnly=completeRingsOnly,
threshold=threshold)
except TypeError:
mcs = MCS.FindMCS(ms,
atomCompare=atomCompare,
bondCompare=bondCompare,
ringMatchesRingOnly=ringMatchesRingOnly,
completeRingsOnly=completeRingsOnly)
res = mcs.smarts
if bool(int(params.get('asSmiles', '0'))):
p = Chem.MolFromSmarts(res)
for m in ms:
if m.HasSubstructMatch(p):
match = m.GetSubstructMatch(p)
res = Chem.MolFragmentToSmiles(m,
atomsToUse=match,
isomericSmiles=True,
canonical=False)
break
return res
#-----------------------------------------------------------------------------------------------------------------------
def test_timeout(self):
t1 = time.time()
result = MCS.FindMCS(lengthy_mols, timeout=0.1)
self.assert_result(result, completed=0)
self.assertTrue(result.numAtoms > 1)
self.assertTrue(result.numBonds >= result.numAtoms-1, (result.numAtoms, result.numBonds))
t2 = time.time()
self.assertTrue(t2-t1 < 0.5, t2-t1)
def make2dcanv(mmpcomps):
"""Function to make a 2D canv from MMPComps"""
rdmols = []
acts = []
subs = []
smsubs = []
donemols = []
for m in mmpcomps:
# Ensure that this is not just the same comparison
mol1 = Chem.MolFromMolBlock((str(m.xtal_mol.sdf_info)))
mol2 = Chem.MolFromMolBlock((str(m.chembl_mol.sdf_info)))
if [m.xtal_mol.cmpd_id.pk, m.chembl_mol.pk] in donemols or [
m.chembl_mol.cmpd_id.pk, m.xtal_mol.pk
] in donemols:
# Don't do the same comparison twice
continue
else:
donemols.append([m.xtal_mol.cmpd_id.pk, m.chembl_mol.cmpd_id.pk])
# Set the molecule name for the 3D display
acts.append(render_act(m.xtal_act))
acts.append(render_act(m.chembl_act))
# Generate the two-d depictions after canonicalising the smiles
mol1 = Chem.MolFromSmiles(Chem.MolToSmiles(mol1, isomericSmiles=True))
mol2 = Chem.MolFromSmiles(Chem.MolToSmiles(mol2, isomericSmiles=True))
smp = MCS.FindMCS([mol1, mol2],
completeRingsOnly=True,
ringMatchesRingOnly=True,
timeout=0.5).smarts
p = Chem.MolFromSmarts(smp)
subs.append(p)
smsubs.append(smp)
AllChem.Compute2DCoords(p)
AllChem.GenerateDepictionMatching2DStructure(mol1,
p,
acceptFailure=True)
AllChem.GenerateDepictionMatching2DStructure(mol2,
p,
acceptFailure=True)
rdmols.extend([mol1, mol2])
# So now we have the mols in a list with actvity information in a list
# Order this list of molecules based on scaffold (p)
# Get a list of the indices of rdmols to rearrange
myinds = sorted(range(len(smsubs)), key=lambda x: smsubs[x])
nmols = []
nacts = []
nsubs = []
# Now rearrange everthing to suit
for ind_m in myinds:
nmols.extend([rdmols[ind_m * 2], rdmols[ind_m * 2 + 1]])
nacts.extend([acts[ind_m * 2], acts[ind_m * 2 + 1]])
nsubs.append(subs[ind_m])
image = draw_acts(nmols, nacts, nsubs)
output = StringIO.StringIO()
image.save(output, format="PNG")
contents = output.getvalue()
return contents
def assert_search(self,
smiles,
numAtoms,
numBonds,
smarts=_ignore,
**kwargs):
result = MCS.FindMCS(smiles, **kwargs)
self.assert_result(result,
completed=1,
numAtoms=numAtoms,
numBonds=numBonds,
smarts=smarts)
def get_decoys(pdb_file, mol_file, num_atoms, init='get_decoys_init'):
"""For each binding ligand, gets a list of decoy ligands. We filter by number
of atoms and maximum common substructure (MCS). Then we generate conformers
for each decoy and save them to the decoy_ligands folder"""
init = eval(init)
reader = SDMolSupplier(mol_file)
mol = reader[0]
output = []
iterator = range(len(init.all_mols))
random.shuffle(iterator)
for i in iterator:
if (init.all_mol_files[i] == mol_file or \
abs(init.all_num_atoms[i] - num_atoms) > init.max_atom_dif):
continue
mcs = MCS.FindMCS([init.all_mols[i], mol],
minNumAtoms=init.max_substruct,
ringMatchesRingOnly=True,
completeRingsOnly=True,
timeout=1)
if mcs.numAtoms == -1:
#save the mol object as a PDB file in the decoys folder
decoy_file = pdb_file.replace('/binding_ligands/',
'/decoy_ligands/').replace(
'.pdb',
str(len(output)) + '.pdb')
pdb_writer = PDBWriter(decoy_file)
# generate the decoy and its conformers
decoy2 = Chem.AddHs(init.all_mols[i])
conf_ids = AllChem.EmbedMultipleConfs(decoy2, init.num_conformers)
for cid in conf_ids:
AllChem.MMFFOptimizeMolecule(decoy2, confId=cid)
decoy = Chem.RemoveHs(decoy2)
pdb_writer.write(decoy)
pdb_writer.close()
output.append([init.all_pdb_files[i], decoy_file])
if len(output) >= init.max_num_decoys:
break
print 'Got the decoys for one ligand'
return output
def get_decoys(pdb_file, mol_file, num_atoms, init='get_decoys_init'):
"""
For each binding ligand, get a list of decoy ligands. We filter by number of atoms and maximum common
substructure (MCS). Returns filepaths to all binding ligand - decoy pair.
:param pdb_file: pdb format ligand
:param mol_file: mol format ligand
:param num_atoms: ligand's atom number
:param init:
:return:
nested list [[pdb_file, decoy_files]]
"""
init = eval(init)
reader = SDMolSupplier(mol_file)
mol = reader[0]
output = ""
counter = 0
# Shuffle which ligands we sample to avoid biases in decoy ligands
iterator = range(len(init.all_mols))
random.shuffle(iterator)
for i in iterator:
if (init.all_mol_files[i] == mol_file
or abs(init.all_num_atoms[i] - num_atoms) >
init.max_atom_dif): # FIXME O2 time
continue # FIXME
mcs = MCS.FindMCS([init.all_mols[i], mol],
minNumAtoms=init.max_substruct,
ringMatchesRingOnly=True,
completeRingsOnly=True,
timeout=1)
if mcs.numAtoms == -1:
if counter == init.max_num_decoys - 1:
output += init.all_pdb_files[i]
counter += 1
break # FIXME
output += init.all_pdb_files[i] + ','
counter += 1
# Check to make sure there are enough decoys
if counter < init.max_num_decoys:
raise Exception("Not enough decoys for ligand " + pdb_file)
print 'Got the decoys for one ligand'
return [[pdb_file, output]]
def create_lexicon(molecule1, molecule2):
#Chem.Kekulize(molecule1)
#Chem.Kekulize(molecule2)
patt1 = Chem.MolFromSmarts(MCS.FindMCS([molecule2, molecule1], matchValences=True).smarts)
matching1 = molecule2.GetSubstructMatch(patt1)
matching2 = molecule1.GetSubstructMatch(patt1)
#below is indices in m, ordered as patt‘s atoms
index1 = range(molecule2.GetNumAtoms())
#these are the atoms in the product that are NOT in the metastructure
product_specific_atoms = list(set(index1) - set(matching1))
matching1 = zip(matching1, range(molecule2.GetNumAtoms()) )
matching2 = zip(matching2, range(molecule1.GetNumAtoms()) )
#lexicon for what values equal what. this is a bit confusing but it's the product's substructure that's similar with the meta-metastructure's
#then the corresponding atom on the meta-metastructure to the metastructure
lexicon = sorted(zip([int(i[0]) for i in matching2], [int(i[0]) for i in matching1]))
return lexicon
def pattern_findersub(steroid1, steroid2, exceptions):
m1 = Chem.MolFromSmiles(steroid1)
m2 = Chem.MolFromSmiles(steroid2)
patt1 = Chem.MolFromSmarts(MCS.FindMCS([Chem.MolFromSmiles(steroid1), Chem.MolFromSmiles(steroid2)]).smarts)
matching1 = m1.GetSubstructMatch(patt1)
matching1 = list(matching1)
####################important exception line
for i in exceptions:
matching1.append( i )
#below is indices in m, ordered as patt‘s atoms
index1 = range(Chem.MolFromSmiles(steroid1).GetNumAtoms())
#these are the atoms in the substrate that are NOT in the product
substrate_specific_atoms = list(set(index1) - set(matching1))
del_bonds = []
add_connections = []
add_bonds = []
del_connections = []
for i in substrate_specific_atoms:
atom = m1.GetAtomWithIdx(i)
#get the bonds that are connected to indexed atom but not the ones that are in the 'meta-structure'
neighbors = [x.GetIdx() for x in atom.GetNeighbors()]
extra_bonds = list(set(neighbors) & set(substrate_specific_atoms))
#get the bonds of these atoms that need to be deleted
bondtype = []
for bond in extra_bonds:
bond = str(m1.GetBondBetweenAtoms(i, bond).GetBondType())
bond = bond.replace('rdkit.Chem.rdchem.BondType.', '')
bondtype.append( bond )
del_connections.append( extra_bonds )
del_bonds.append( bondtype )
substrate_modifications = pd.DataFrame({'Substrate Unique Atoms': substrate_specific_atoms, 'Connections to be deleted': del_connections, 'Bonds to be deleted': del_bonds})
return substrate_modifications
def moonshot():
from dgllife.utils import mol_to_bigraph, CanonicalAtomFeaturizer
import pandas as pd
import os
df = pd.read_csv(
os.path.dirname(graca.data.collections.__file__) +
"/covid_submissions_all_info.csv")
df = df.dropna(subset=["f_avg_pIC50"])
from rdkit import Chem
from rdkit.Chem import MCS
ds = []
for idx0, row0 in df.iterrows():
smiles0 = row0["SMILES"]
mol0 = Chem.MolFromSmiles(smiles0)
for idx1, row1 in df.iloc[idx0 + 1:].iterrows():
smiles1 = row1["SMILES"]
mol1 = Chem.MolFromSmiles(smiles1)
res = MCS.FindMCS([mol0, mol1])
if res.numAtoms > 15:
ds.append((
mol_to_bigraph(mol1,
node_featurizer=CanonicalAtomFeaturizer(
atom_data_field='feat')),
mol_to_bigraph(mol0,
node_featurizer=CanonicalAtomFeaturizer(
atom_data_field='feat')),
row1["f_avg_pIC50"],
row0["f_avg_pIC50"],
))
ds_tr = ds[:500]
ds_te = ds[500:]
return ds_tr, ds_te
def modify_metastructure(product_modifications, metastructure, steroid2):
m2 = Chem.MolFromSmiles(steroid2)
#msubstrate = Chem.MolFromSmiles(steroid1)
patt1 = Chem.MolFromSmarts(MCS.FindMCS([metastructure, m2], matchValences=True).smarts)
#convert number of product specifc atom to our metastructure
anchors = [] #anchors are in the MCS, they will ultimately be deleted but are important for figuring out where to add bonds
anchortype = []
lexicon = create_lexicon(metastructure, m2)
for i in product_modifications['Connections to be added'].tolist():
for k in i: #connections in list if there are multiple
for j in lexicon:
if k == j[1]:
anchors.append( j[0] )
atom = metastructure.GetAtomWithIdx(int(j[0]))
anchortype.append( atom.GetAtomicNum())
neighbors = [] #for every index gives the neighbors of the same index in the pandas DF earlier
for i in anchors:
adjacent_atoms = []
atom = metastructure.GetAtomWithIdx(i)
adjacent_atoms = [x.GetIdx() for x in atom.GetNeighbors()]
neighbors.append( adjacent_atoms )
#add the product-specific atoms
em = Chem.EditableMol(metastructure)
newindexes = []
newanchors = []
###KEEP THESE DELETED####
'''for i in product_modifications['Atomic Number'].tolist():
newidx = em.AddAtom(Chem.Atom( int(i) ))
newindexes.append( newidx )
for i in range(len(anchors)):
newanchor = em.AddAtom(Chem.Atom( anchortype[i] ))
newanchors.append( newanchor )'''
#####logic gate for if a carboxyl like addition is going on
similar_indices = []
for i in range(len(product_modifications)):
for j in range(len(product_modifications)):
if product_modifications['Connections to be added'].irow(i) == product_modifications['Connections to be added'].irow(j):
similar_indices.append( i )
if len(similar_indices) > 2:
#translate neighbor number to what it corresponds to in m1
m1 = em.GetMol()
for i in lexicon:
if product_modifications['Neighbors'].irow( similar_indices[0] )[0] == i[1]:
neighbor = i[0]
atom = m1.GetAtomWithIdx(neighbor)
neighbortype = atom.GetAtomicNum()
neighbor_of_neighbor = [x.GetIdx() for x in atom.GetNeighbors()]
neighboranchor = em.AddAtom(Chem.Atom( int(neighbortype) ))
new_atoms = []
for i in range(len(product_modifications)):
new_atom = em.AddAtom( Chem.Atom( int(product_modifications['Atomic Number'].irow(i)) ))
new_atoms.append( new_atom )
for i in range(len(product_modifications)):
if str(product_modifications['Bonds to be added'].irow(i)[0]) == 'DOUBLE':
em.AddBond( int(neighboranchor),int(new_atoms[i]), Chem.BondType.DOUBLE)
elif str(product_modifications['Bonds to be added'].irow(i)[0]) == 'SINGLE':
em.AddBond(int(neighboranchor),int(new_atoms[i]), Chem.BondType.SINGLE)
for i in neighbor_of_neighbor:
em.AddBond(int(i), int(neighboranchor), Chem.BondType.SINGLE)
#get rid of old anchor
for i in list(set(anchors)):
em.RemoveAtom(i)
else:
#add the product-specific atoms
em = Chem.EditableMol(metastructure)
newindexes = []
newanchors = []
for i in product_modifications['Atomic Number'].tolist():
newidx = em.AddAtom(Chem.Atom( int(i) ))
newindexes.append( newidx )
for i in range(len(anchors)):
newanchor = em.AddAtom(Chem.Atom( anchortype[i] ))
newanchors.append( newanchor )
mref = em.GetMol()
#combine the new atom with it's new anchor
for i in range(len(newindexes)):
if str(product_modifications['Bonds to be added'][i][0]) == 'DOUBLE':
em.AddBond(int(newindexes[i]), int(newanchors[i]), Chem.BondType.DOUBLE)
elif str(product_modifications['Bonds to be added'][i][0]) == 'SINGLE':
em.AddBond(int(newindexes[i]), int(newanchors[i]), Chem.BondType.SINGLE)
#combine new structure (newanchor + new atom) to the neighbors of the old anchor
for i in range(len(anchors)):
for j in range(len(neighbors[i])):
atom = mref.GetAtomWithIdx( int(neighbors[i][j]) )
em.AddBond(int(newanchors[i]), int(neighbors[i][j]), Chem.BondType.SINGLE)
#get rid of old anchor
for i in anchors:
em.RemoveAtom(i)
m1 = em.GetMol()
for atom in m1.GetAtoms():
atom.SetNumRadicalElectrons(0)
Chem.SanitizeMol(m1)
return m1
def modify_substrate(substrate_modifications, steroid1, steroid2, steroids):
m1 = Chem.MolFromSmiles(steroid1)
m2 = Chem.MolFromSmiles(steroid2)
ms = Chem.MolFromSmiles(steroids)
############################################################
#removes atoms that are removed via lyase activity, first must find atoms that are removed from the native substrate to the product
#then we have to compare those atoms to our non-native substrate then systematically remove them the tricky thing here will be indexing (as always)
native_lexicon = create_lexicon(ms, m2)
substrates_lexicon = create_lexicon(m1, ms)
ms_atoms = []
ms_m2_atoms = []
m2matchingatoms = []
for atom in ms.GetAtoms():
ms_atoms.append( atom.GetIdx() )
for i in native_lexicon:
ms_m2_atoms.append( i[0] )
m2matchingatoms.append( i[1] )
#see which atoms don't have overlap i.e. things that need to be deleted
ms_cleaved = []
for i in ms_atoms:
if i not in ms_m2_atoms:
ms_cleaved.append( i )
mp_atoms = []
for atom in m2.GetAtoms():
mp_atoms.append( atom )
mp_unique_atoms = []
for i in mp_atoms:
if i.GetIdx() not in m2matchingatoms:
mp_unique_atoms.append( i.GetIdx() )
#aromatic atoms will screw this code up, we need to make sure the atoms we're going to delete are due to aromaticitiy
aromatic_check = []
aromatic_idxi = []
aromatic_idxj = []
for i in mp_unique_atoms:
for j in mp_unique_atoms:
try:
aromatic_check.append( str( m2.GetBondBetweenAtoms(i, j).GetIsAromatic() ) )
aromatic_idxi.append( i )
aromatic_idxj.append( j )
except:
pass
#translate ms_cleaved to our target
target_specifics = []
for i in ms_cleaved:
for j in substrates_lexicon:
if i == j[1]:
target_specifics.append( j[0] )
if 'True' not in aromatic_check:
AROMATIC_FLAG = None
temp_lex = create_lexicon( ms, m1 )
em = Chem.EditableMol(m1)
for i in range(len(ms_cleaved)):
temp_lex = create_lexicon( ms, m1 )
for j in temp_lex:
if j[0] == ms_cleaved[i]:
deletion_atom = j[1]
em.RemoveAtom(deletion_atom)
m1 = em.GetMol()
m1smiles = Chem.MolToSmiles( m1 )
m1smiles = clean_smiles( m1smiles )
try:
m1 = Chem.MolFromSmiles( m1smiles )
except:
pass
else:
AROMATIC_FLAG = 'GO'
#create lexicon to compare atom indices'''
lexicon = create_lexicon(m1, m2)
############################################################
#This will be the double bonds specific to the substrate
#I have to do this because double bonds seem to be more specific than single bonds in RDkit,
#By knowing the exact position of the double bonds I need to remove and add, I can more accurately transform the molecule
m2bondtypessub = []
m2bondidxsub = []
m2bondstartsub = []
m2bondendsub = []
m1bondtypessub = []
m1bondidxsub = []
m1bondstartsub = []
m1bondendsub = []
for i in lexicon:
idx1 = i[0]
m1bondidxsub.append( idx1 )
m1bondtypessub.append( m1.GetBondWithIdx(idx1).GetBondType() )
for i in m1bondidxsub:
m1bondstartsub.append( m1.GetBondWithIdx(int(i)).GetBeginAtomIdx() )
m1bondendsub.append( m1.GetBondWithIdx(int(i)).GetEndAtomIdx() )
for i in m1bondstartsub:
for j in lexicon:
if i == j[0]:
m2bondstartsub.append( j[1] )
for i in m1bondendsub:
for j in lexicon:
if i == j[0]:
m2bondendsub.append( j[1] )
for i in m1bondidxsub:
for j in lexicon:
if i == j[0]:
m2bondidxsub.append( j[1] )
for i in m2bondidxsub:
m2bondtypessub.append( m2.GetBondWithIdx(int(i)).GetBondType() )
bondindices = ''
bondindicessub = ''
em = Chem.EditableMol(m1)
if len(BondIndex( ms, m2 )) != 0 and AROMATIC_FLAG != 'GO': #compares bonds between both the substrate, product, and target to determine if anything needs to be added
bondindicessub = BondIndex( m1, m2 )
bondindicessub = bondindicessub[bondindicessub['M1 Bond Type'] != bondindicessub['M2 Bond Type']]
#bondindices = bondindices[bondindices['M1 Bond Index'] != bondindices['M1 Bond Start']]
em = Chem.EditableMol(m1)
if len(BondIndex(m1, m2)) != 0: ###This logic gate passes the bond formation if the substrate > product chemistry has no bond changes. I may actually destroy this step entirely...
for i in range(len(bondindicessub)):
if str(bondindicessub['M1 Bond Type'].irow(i)) == str('SINGLE') and str(bondindicessub['M2 Bond Type'].irow(i)) == str('DOUBLE'):
em.RemoveBond(int(bondindicessub['M1 Bond Start'].irow(i)), int(bondindicessub['M1 Bond End'].irow(i)))
em.AddBond(int(bondindicessub['M1 Bond Start'].irow(i)), int(bondindicessub['M1 Bond End'].irow(i)), Chem.BondType.DOUBLE)
else:
str(bondindicessub['M1 Bond Type'].irow(i)) == str('DOUBLE')
em.RemoveBond(int(bondindicessub['M1 Bond Start'].irow(i)), int(bondindicessub['M1 Bond End'].irow(i)))
em.AddBond(int(bondindicessub['M1 Bond Start'].irow(i)), int(bondindicessub['M1 Bond End'].irow(i)), Chem.BondType.SINGLE)
else:
pass
m1 = em.GetMol()
############################################################iterate through bonds in both molecules to see if we need to delete any
#This will be the double bonds specific to the product
m2bondtypes = []
m2bondidx = []
m2bondstart = []
m2bondend = []
m1bondtypes = []
m1bondidx = []
m1bondstart = []
m1bondend = []
for i in lexicon:
idx2 = i[1]
m2bondidx.append( idx2 )
m2bondtypes.append( m2.GetBondWithIdx(idx2).GetBondType() )
for i in m2bondidx:
m2bondstart.append( m2.GetBondWithIdx(int(i)).GetBeginAtomIdx() )
m2bondend.append( m2.GetBondWithIdx(int(i)).GetEndAtomIdx() )
m1bondtypes = []
m1bondidx = []
m1bondstart = []
m1bondend = []
for i in m2bondstart:
for j in lexicon:
if i == j[1]:
m1bondstart.append( j[0] )
for i in m2bondend:
for j in lexicon:
if i == j[1]:
m1bondend.append( j[0] )
for i in m2bondidx:
for j in lexicon:
if i == j[1]:
m1bondidx.append( j[0] )
for i in m1bondidx:
m1bondtypes.append( m1.GetBondWithIdx(int(i)).GetBondType() )
#exceptions are atoms that extend from the start bond that actually are irrelevant because where they should go don't exist in the starting molecule
#this is problematic because they could add bonds to places they shouldn't be
exceptions = []
for i in range(len(m2bondend)):
if m2bondend[i] not in [int(j[1]) for j in lexicon]:
exceptions.append( m2bondend[i] )
try: #this try skips the deconstruction phase if it doesn't need to happen
bondindices = BondIndex( m2, m1 )
patt1 = Chem.MolFromSmarts(MCS.FindMCS([Chem.MolFromSmiles(steroid1), Chem.MolFromSmiles(steroid2)]).smarts)
#remove specififed atoms from the substrate
to_delete_atoms = substrate_modifications['Substrate Unique Atoms'].tolist()
#deletes atoms that are in the substrate but not the product, could be from lyases or whatnot
for i in range(len(substrate_modifications['Substrate Unique Atoms'])):
try:
patt1 = Chem.MolFromSmarts(MCS.FindMCS([m1, Chem.MolFromSmiles(steroid2)]).smarts)
matching1 = m1.GetSubstructMatch(patt1)
matching1 = list(matching1)
#below is indices in m, ordered as patt‘s atoms
index1 = range(m1.GetNumAtoms())
#these are the atoms in the substrate that are NOT in the product
substrate_specific_atoms = list(set(index1) - set(matching1))
#delete just the first substrate_specific_atom because the whole molecule will reindex
em = Chem.EditableMol(m1)
em.RemoveAtom(to_delete_atoms[0])
#need to fix the valences of the atoms we deleted earlier, for whatever reason a radical or a hydrogen is thrown on and the valence is incorrect.
#Using bondindicessub because it's coordinates are 100% reliable
for j in range(len(bondindicessub)):
if int(bondindicessub['M1 Bond End'].irow(j)) == int(substrate_specific_atoms[0]):
deletion_target_num = int(bondindicessub['M1 Bond Start'].irow(j))
deletion_target = m1.GetAtomWithIdx( int(bondindicessub['M1 Bond Start'].irow(j)) )
deletion_neighbors = [x.GetIdx() for x in deletion_target.GetNeighbors()]
#get the bonds that are connected to indexed atom but not the ones that are in the 'meta-structure'
em = Chem.EditableMol(m1)
newidx = em.AddAtom(Chem.Atom(6))
if len(deletion_neighbors) == 1:
em.AddBond(newidx, deletion_neighbors[0], Chem.BondType.SINGLE)
elif len(deletion_neighbors) == 2:
em.AddBond(newidx, deletion_neighbors[0], Chem.BondType.SINGLE)
em.AddBond(newidx, deletion_neighbors[1], Chem.BondType.SINGLE)
elif len(deletion_neighbors) == 3:
em.AddBond(newidx, deletion_neighbors[0], Chem.BondType.SINGLE)
em.AddBond(newidx, deletion_neighbors[1], Chem.BondType.SINGLE)
em.AddBond(newidx, deletion_neighbors[2], Chem.BondType.SINGLE)
em.RemoveAtom(deletion_target_num)
except:
pass
except:
bondindices = ''
bondindicessub = ''
m1 = em.GetMol()
#Chem.SanitizeMol(m1)
return m1, bondindices, bondindicessub, AROMATIC_FLAG
def do_lloommppaa_proc(target_id,
pdb_protein,
smiles,
mol2_protein=None,
reactants=None,
products=None,
context=None):
"""Function to do the processing for LLOOMMPPAA"""
from LLOOMMPPAA.models import PossReact
from LLOOMMPPAA.reactions import run_react_proc, define_reacts, load_in_reagents, load_in_follow_ups, find_follow_ups, define_reaction_process
# Load the data
load_wonka_prot(target_id, pdb_protein, smiles)
if reactants:
# Find the potential synthesis vectors
define_reacts()
find_follow_ups(target_id=target_id)
# Show the sides -> USER MUST SELECT A SIDE
poss_reacts = PossReact.objects.filter(
mol_id__prot_id__target_id=target_id)
# Loop through all the possible reactions
for ps_r in poss_reacts:
print ps_r.react_id.name
print "1):", ps_r.replaced_frag
print "2):", ps_r.retained_frag
print "3):", "SKIP"
choice = int(raw_input("Select a fragment to replace..."))
if choice == 1:
context = ps_r.retained_frag_context
react_frag = ps_r.retained_frag
this_react = ps_r.react_id
break
elif choice == 2:
context = ps_r.replaced_frag_context
react_frag = ps_r.replaced_frag
this_react = ps_r.react_id
break
else:
continue
if not context:
print "YOU MUST SPECIFY A CONTEXT"
if products:
print "AUTO GENERATING FROM PRODUCTS"
print products
context = Chem.CanonSmiles(
Chem.MolToSmiles(Chem.MolFromSmarts(
MCS.FindMCS(Chem.SDMolSupplier(products)).smarts),
isomericSmiles=True))
print context
else:
return
if reactants:
mol_id = ps_r.mol_id
prot_id = ps_r.mol_id.prot_id
else:
mol_id = Molecule.objects.filter(prot_id__target_id=target_id,
smiles=smiles)[0]
prot_id = mol_id.prot_id
my_prots = [prot_id]
# Set the mol2 protein for this target -> throw a warning if this doesn't happen
if mol2_protein:
from PLIFS.models import PlifProtein
pp = PlifProtein()
pp.prot_id = prot_id
pp.mol2_data = open(mol2_protein).read()
pp.save()
# Define the reactants and products
if reactants:
react_id = load_in_reagents("RUN_DEF", reactants, ps_r.react_id)
else:
react_id = None
if products:
my_react = Reaction.objects.get_or_create(name="DUMMY",
react_smarts="DUMMY",
retro_smarts="DUMMY",
cont_smarts="DUMMY")[0]
this_react = my_react
react_frag = context
prod_id = load_in_follow_ups("RUN_PROD", products, my_react, mol_id)
else:
prod_id = None
# Now set up the reaction itself
react_proc = define_reaction_process(mol_id,
context,
my_prots,
this_react,
context,
react_frag,
products_id=prod_id,
reactants_id=react_id)
if react_id:
react_proc.proc_stage = "RUN_REACTION"
react_proc.save()
if products:
react_proc.proc_stage = "GENERATE CONFS"
react_proc.save()
# Now run the reaction and analysis itself
run_react_proc(react_proc)
# xyz to Mol
## call babel
## babel -ixyz
query_mol = readfile(file1)
template_mol = readfile(file2)
# mcs
skeleton_mcs = MCS.FindMCS([query_mol, template_mol])
skeleton_mol = Chem.MolFromSmarts(skeleton_mcs.smarts)
min_overlap = 6 # Require the overlap to be at least min_overlap atoms
if ( len(skeleton_mol.GetAtoms()) < min_overlap):
print("These molecules share less than min_overlap atoms. Quitting.")
query_mcs_matches = query_mol.GetSubstructMatches(skeleton_mol)
print("query_mol contains overlapping fragment " + len(query_mcs_matches) + " times.")
template_mcs_matches = template_mol.GetSubstructMatches(skeleton_mol)
print("template_mol contains overlapping fragment " + len(template_mcs_matches) + " times.")
for query_mcs_match_index in range(len(query_mcs_matches)):
if (arg[2].endswith('.smi')):
optSuppl = Chem.SmilesMolSupplier(arg[2])
elif (arg[2].endswith('sdf')):
optSuppl = Chem.SDMolSupplier(arg[2])
else:
print('File type not supported')
sys.exit()
w = Chem.SDWriter('alignLeadsOut.sdf')
# Use align mol
# Get common substructure for lead and .smi
# Align copy to lead
# Use AlignMolConformers with argument to specific atom IDs of substructure
mols = [lead, optSuppl[0]]
res = MCS.FindMCS(
mols, threshold=0.9, completeRingsOnly=True
) # Calculates most common substructure and outputs SMARTS pattern
p = Chem.MolFromSmarts(
res.smarts) # Creates mol object of most common substructure
core = AllChem.DeleteSubstructs(
AllChem.ReplaceSidechains(Chem.RemoveHs(lead), p),
Chem.MolFromSmiles('*'))
core.UpdatePropertyCache()
for mol in optSuppl:
try:
AllChem.ConstrainedEmbed(mol, core, useTethers=False)
w.write(mol)
except:
pass
def fmcstimeout(p, q):
return MCS.FindMCS([p, q], timeout=0.01).smarts
def fmcstimeout(p, q):
return MCS.FindMCS([p, q]).smarts
def view_2dmol(option,
maps=None,
out_put=None,
target_id=None,
legends=None,
extra=None):
"""Function to render a mol image from a smiles.
The input (option) could be 1) a list of smiles 2) a smiles 3) pdb_code
Returns a molecule image as data"""
option = str(option)
print option
try:
option = ast.literal_eval(option)
except:
pass
if type(option) is list:
mols = [Chem.MolFromSmiles(str(x)) for x in option]
p = Chem.MolFromSmarts(MCS.FindMCS(mols).smarts)
AllChem.Compute2DCoords(p)
[AllChem.GenerateDepictionMatching2DStructure(x, p) for x in mols]
image = Draw.MolsToGridImage(mols, 2, legends=legends)
# If it's a PDB code
elif Chem.MolFromSmiles(str(option)) is None and type(option) is str:
mol = Chem.MolFromSmiles(
str(
Molecule.objects.filter(
prot_id__code=option)[0].cmpd_id.smiles))
AllChem.GenerateDepictionMatching3DStructure(
mol,
Chem.MolFromMolBlock(
str(Molecule.objects.filter(
prot_id__code=option)[0].sdf_info)))
image = Draw.MolToImage(mol)
# If it's a
elif type(option) is str:
mol = Chem.MolFromSmiles(str(option))
if extra == "SIMPLE":
image = Draw.MolToImage(mol)
elif extra is None:
h_map = None
image = Draw.MolToImage(mol,
size=(100, 100),
fitImage=True,
highlightMap=h_map)
else:
sub = Chem.MolFromSmiles(str(extra))
h_map = get_h_map(mol, sub)
image = Draw.MolToImage(mol, highlightMap=h_map)
if maps is None:
pass
else:
maps = float(maps)
draw = ImageDraw.Draw(image)
dim = (20, 0) + (20, image.size[1])
draw.line(dim,
fill=(255 - int(255 * maps), int(255 * maps), 0),
width=10)
else:
print "NOT VALID TYPE"
return "NOT VALID TYPE"
output = StringIO.StringIO()
image.save(output, format="PNG")
contents = output.getvalue()
return contents