def substitutions(rct_gras, prd_gras):
""" find substitutions consistent with these reactants and products
:param rct_gras: reactant graphs (must have non-overlapping keys)
:param prd_gras: product graphs (must have non-overlapping keys)
Substitutions are identified by breaking one bond in the reactants and one
bond from the products and checking for isomorphism.
"""
_assert_is_valid_reagent_graph_list(rct_gras)
_assert_is_valid_reagent_graph_list(prd_gras)
rxns = []
if len(rct_gras) == 2 and len(prd_gras) == 2:
rct_gra = union_from_sequence(rct_gras)
prd_gra = union_from_sequence(prd_gras)
for rgra1, rgra2 in itertools.permutations(rct_gras):
bnd_keys = bond_keys(rgra1)
rad_keys = unsaturated_atom_keys(rgra2)
for bnd_key, rad_key in itertools.product(bnd_keys, rad_keys):
gra = remove_bonds(rct_gra, [bnd_key])
for brk_key1 in bnd_key:
gra = add_bonds(gra, [(brk_key1, rad_key)])
inv_dct = isomorphism(gra, prd_gra)
if inv_dct:
brk_key2, = bnd_key - {brk_key1}
f_frm_bnd_key = (brk_key1, rad_key)
f_brk_bnd_key = (brk_key1, brk_key2)
b_frm_bnd_key = (inv_dct[brk_key1], inv_dct[brk_key2])
b_brk_bnd_key = (inv_dct[brk_key1], inv_dct[rad_key])
forw_tsg = ts.graph(rct_gra,
frm_bnd_keys=[f_frm_bnd_key],
brk_bnd_keys=[f_brk_bnd_key])
back_tsg = ts.graph(prd_gra,
frm_bnd_keys=[b_frm_bnd_key],
brk_bnd_keys=[b_brk_bnd_key])
rcts_atm_keys = [atom_keys(rgra1), atom_keys(rgra2)]
prds_atm_keys = list(map(atom_keys, prd_gras))
if inv_dct[rad_key] not in prds_atm_keys[0]:
prds_atm_keys = list(reversed(prds_atm_keys))
# Create the reaction object
rxns.append(
Reaction(
rxn_cls=par.ReactionClass.SUBSTITUTION,
forw_tsg=forw_tsg,
back_tsg=back_tsg,
rcts_keys=rcts_atm_keys,
prds_keys=prds_atm_keys,
))
return ts_unique(rxns)
def ring_forming_scissions(rct_gras, prd_gras):
""" find ring-forming scissions consistent with these reactants and products
:param rct_gras: reactant graphs (must have non-overlapping keys)
:param prd_gras: product graphs (must have non-overlapping keys)
Ring-forming scissions are found by breaking ring-bonds on one product and
joining the ends to unsaturated sites on the other product
"""
_assert_is_valid_reagent_graph_list(rct_gras)
_assert_is_valid_reagent_graph_list(prd_gras)
rxns = []
if len(rct_gras) == 1 and len(prd_gras) == 2:
rgra, = rct_gras
pgra = union_from_sequence(prd_gras)
for pgra1, pgra2 in itertools.permutations(prd_gras):
bnd_keys = list(itertools.chain(*rings_bond_keys(pgra1)))
atm_keys = unsaturated_atom_keys(pgra2)
for bnd_key, atm_key in itertools.product(bnd_keys, atm_keys):
# Break a ring bond
gra = remove_bonds(pgra, [bnd_key])
for end_key in bnd_key:
# Add to one end of the broken ring
fgra = add_bonds(gra, [(atm_key, end_key)])
inv_dct = isomorphism(fgra, rgra)
if inv_dct:
other_end_key, = bnd_key - {end_key}
f_frm_bnd_key = (inv_dct[end_key],
inv_dct[other_end_key])
f_brk_bnd_key = (inv_dct[end_key], inv_dct[atm_key])
b_frm_bnd_key = (end_key, atm_key)
b_brk_bnd_key = (end_key, other_end_key)
forw_tsg = ts.graph(rgra,
frm_bnd_keys=[f_frm_bnd_key],
brk_bnd_keys=[f_brk_bnd_key])
back_tsg = ts.graph(pgra,
frm_bnd_keys=[b_frm_bnd_key],
brk_bnd_keys=[b_brk_bnd_key])
# Create the reaction object
rxns.append(
Reaction(
rxn_cls=par.ReactionClass.RING_FORM_SCISSION,
forw_tsg=forw_tsg,
back_tsg=back_tsg,
rcts_keys=[atom_keys(rgra)],
prds_keys=[atom_keys(pgra1),
atom_keys(pgra2)],
))
return ts_unique(rxns)
def test__vmat__vmatrix():
""" test graph.vmat.vmatrix
"""
ich = automol.smiles.inchi('C12CC(C1)C2CC3C(C3)CCC4C5CCC(CC5)C4')
gra = automol.inchi.graph(ich)
_, zma_keys = graph.vmat.vmatrix(gra)
assert set(zma_keys) == graph.atom_keys(gra)
def _geometry_from_info(gra, rct_geos, geo_init, dist_range_dct,
relax_ang=False, relax_tors=False,
pla_dct=None,
max_dist_err=2e-1, log=False):
keys = sorted(atom_keys(gra))
xmat = automol.geom.coordinates(geo_init, angstrom=True)
lmat, umat = automol.graph.embed.join_distance_bounds_matrices(
gra, keys, dist_range_dct, geos=rct_geos, relax_angles=relax_ang,
relax_torsions=relax_tors)
pla_dct = {} if pla_dct is None else pla_dct
chi_dct = automol.graph.embed.chirality_constraint_bounds(gra, keys)
pla_dct.update(automol.graph.embed.planarity_constraint_bounds(gra, keys))
xmat, conv = automol.embed.cleaned_up_coordinates(
xmat, lmat, umat, chi_dct=chi_dct, pla_dct=pla_dct,
max_dist_err=max_dist_err, log=log)
if log:
print("Converged!" if conv else "Did not converge.")
syms = list(itertools.chain(*map(automol.geom.symbols, rct_geos)))
xyzs = xmat[:, :3]
geo = automol.geom.from_data(syms, xyzs, angstrom=True)
return geo
def test__set_atom_implicit_hydrogen_valences():
""" test graph.set_atom_implicit_hydrogen_valences
"""
atm_keys = graph.atom_keys(C8H13O_CGR)
cgr = graph.set_atom_implicit_hydrogen_valences(
C8H13O_CGR, {atm_key: 0
for atm_key in atm_keys})
assert cgr == automol.graph.from_data(graph.atom_symbols(C8H13O_CGR),
graph.bond_keys(C8H13O_CGR))
def _partial_hydrogen_abstraction(qh_gra, q_gra):
rets = []
h_atm_key = max(atom_keys(q_gra)) + 1
uns_atm_keys = unsaturated_atom_keys(q_gra)
for atm_key in uns_atm_keys:
q_gra_h = add_atom_explicit_hydrogen_keys(q_gra,
{atm_key: [h_atm_key]})
inv_atm_key_dct = isomorphism(q_gra_h, qh_gra)
if inv_atm_key_dct:
qh_q_atm_key = inv_atm_key_dct[atm_key]
qh_h_atm_key = inv_atm_key_dct[h_atm_key]
q_q_atm_key = atm_key
rets.append((qh_q_atm_key, qh_h_atm_key, q_q_atm_key))
return rets
def hydrogen_abstractions(rct_gras, viable_only=True):
""" find hydrogen abstraction reactions for these reactants
:param rct_gras: graphs for the reactants, without stereo and without
overlapping keys
:param viable_only: Filter out reactions with non-viable products?
:type viable_only: bool
:returns: a list of Reaction objects
:rtype: tuple[Reaction]
Hydrogen abstractions are enumerated by looping over unique unsaturated
atoms on one molecule and abstracting from unique atoms on the other.
"""
assert_is_valid_reagent_graph_list(rct_gras)
rxns = []
if len(rct_gras) == 2:
for q1h_gra, q2_gra in itertools.permutations(rct_gras):
hyd_keys = atom_keys(q1h_gra, sym='H')
# Identify unique heavy atoms as potential donors
don_keys = atom_keys(q1h_gra, excl_syms=('H', ))
don_keys = atom_equivalence_class_reps(q1h_gra, don_keys)
# Identify unique unsaturated atoms as potential attackers
att_keys = unsaturated_atom_keys(q2_gra)
att_keys = atom_equivalence_class_reps(q2_gra, att_keys)
for don_key, att_key in itertools.product(don_keys, att_keys):
hyd_key = atom_neighbor_atom_key(q1h_gra,
don_key,
symbs_first=['H'],
symbs_last=[])
if hyd_key in hyd_keys:
# Remove a hydrogen from the donor site
q1_gra = remove_atoms(q1h_gra, {hyd_key})
# Add a hydrogen atom to the attacker site
q2h_gra = add_bonded_atom(q2_gra,
'H',
att_key,
bnd_atm_key=hyd_key)
rcts_gra = union(q1h_gra, q2_gra)
prds_gra = union(q2h_gra, q1_gra)
forw_tsg = ts.graph(rcts_gra,
frm_bnd_keys=[(att_key, hyd_key)],
brk_bnd_keys=[(don_key, hyd_key)])
back_tsg = ts.graph(prds_gra,
frm_bnd_keys=[(don_key, hyd_key)],
brk_bnd_keys=[(att_key, hyd_key)])
rcts_atm_keys = list(map(atom_keys, [q1h_gra, q2_gra]))
prds_atm_keys = list(map(atom_keys, [q2h_gra, q1_gra]))
# Create the reaction object
rxns.append(
Reaction(
rxn_cls=par.ReactionClass.Typ.HYDROGEN_ABSTRACTION,
forw_tsg=forw_tsg,
back_tsg=back_tsg,
rcts_keys=rcts_atm_keys,
prds_keys=prds_atm_keys,
))
if viable_only:
rxns = filter_viable_reactions(rxns)
return ts_unique(rxns)
def hydrogen_migrations(rct_gras, viable_only=True):
""" find all possible hydrogen migration reactions for these reactants
:param rct_gras: graphs for the reactants, without stereo and without
overlapping keys
:param viable_only: Filter out reactions with non-viable products?
:type viable_only: bool
:returns: a list of Reaction objects
:rtype: tuple[Reaction]
Hydrogen migrations are enumerated looping over unsaturated sites, adding
hydrogens to them, and looping over non-equivalent heavy atoms and removing
hydrgens from them.
"""
assert_is_valid_reagent_graph_list(rct_gras)
rxns = []
if len(rct_gras) == 1:
rct_gra, = rct_gras
# Identify unsaturated sites
rct_add_key = max(atom_keys(rct_gra)) + 1
rct_rad_keys = unsaturated_atom_keys(rct_gra)
rct_hyd_keys = atom_keys(rct_gra, sym='H')
for rct_rad_key in rct_rad_keys:
# Add a hydrogen to the radical/unsaturated site
rct_h_gra = add_bonded_atom(rct_gra,
'H',
rct_rad_key,
bnd_atm_key=rct_add_key)
# Identify donor sites
rct_don_keys = backbone_keys(rct_h_gra) - {rct_rad_key}
for rct_don_key in rct_don_keys:
rct_hyd_key = atom_neighbor_atom_key(rct_gra,
rct_don_key,
symbs_first=['H'],
symbs_last=[])
if rct_hyd_key in rct_hyd_keys:
prd_gra = remove_atoms(rct_h_gra, {rct_hyd_key})
prd_gra = relabel(prd_gra, {rct_add_key: rct_hyd_key})
forw_tsg = ts.graph(rct_gra,
frm_bnd_keys=[(rct_rad_key,
rct_hyd_key)],
brk_bnd_keys=[(rct_don_key,
rct_hyd_key)])
back_tsg = ts.graph(prd_gra,
frm_bnd_keys=[(rct_don_key,
rct_hyd_key)],
brk_bnd_keys=[(rct_rad_key,
rct_hyd_key)])
rxns.append(
Reaction(
rxn_cls=par.ReactionClass.Typ.HYDROGEN_MIGRATION,
forw_tsg=forw_tsg,
back_tsg=back_tsg,
rcts_keys=[atom_keys(rct_gra)],
prds_keys=[atom_keys(prd_gra)],
))
if viable_only:
rxns = filter_viable_reactions(rxns)
return ts_unique(rxns)
def eliminations(rct_gras, viable_only=True):
""" find all possible elimination reactions for these reactants
:param rct_gras: graphs for the reactants, without stereo and without
overlapping keys
:param viable_only: Filter out reactions with non-viable products?
:type viable_only: bool
:returns: a list of Reaction objects
:rtype: tuple[Reaction]
Eliminations are enumerated by forming a bond between an attacking heavy
atom and another atom not initially bonded to it, forming a ring. The bond
adjacent to the attacked atom is then broken, along with a second bond in
the ring, downstream from the attacking heavy atom, away from the attacked
atom.
"""
assert_is_valid_reagent_graph_list(rct_gras)
rxns = []
if len(rct_gras) == 1:
rct_gra, = rct_gras
ngb_keys_dct = atoms_neighbor_atom_keys(rct_gra)
# frm1_keys = atom_keys(rct_gra, excl_syms=('H',))
frm1_keys = unsaturated_atom_keys(rct_gra)
rct_symbs = atom_symbols(rct_gra)
frm1_keys_o = frozenset(key for key in frm1_keys
if rct_symbs[key] == 'O')
frm2_keys = atom_keys(rct_gra)
bnd_keys = bond_keys(rct_gra)
frm_bnd_keys = [(frm1_key, frm2_key)
for frm1_key, frm2_key in itertools.product(
frm1_keys_o, frm2_keys) if frm1_key != frm2_key
and not frozenset({frm1_key, frm2_key}) in bnd_keys]
for frm1_key, frm2_key in frm_bnd_keys:
# Bond the radical atom to the hydrogen atom
prds_gra = add_bonds(rct_gra, [(frm2_key, frm1_key)])
# Get keys to the ring formed by this extra bond
rng_keys = next((ks for ks in rings_atom_keys(prds_gra)
if frm2_key in ks and frm1_key in ks), None)
# Eliminations (as far as I can tell) only happen through TSs with
# 3- or 4-membered rings
if rng_keys is not None and len(rng_keys) < 5:
frm1_ngb_key, = ngb_keys_dct[frm1_key] & set(rng_keys)
frm2_ngb_key, = ngb_keys_dct[frm2_key] & set(rng_keys)
# Break the bonds on either side of the newly formed bond
prds_gra = remove_bonds(prds_gra, [(frm1_key, frm1_ngb_key)])
prds_gra = remove_bonds(prds_gra, [(frm2_key, frm2_ngb_key)])
prd_gras = connected_components(prds_gra)
if len(prd_gras) == 2:
forw_tsg = ts.graph(rct_gra,
frm_bnd_keys=[(frm1_key, frm2_key)],
brk_bnd_keys=[(frm1_key, frm1_ngb_key),
(frm2_key, frm2_ngb_key)
])
back_tsg = ts.graph(prds_gra,
frm_bnd_keys=[(frm1_key, frm1_ngb_key),
(frm2_key, frm2_ngb_key)
],
brk_bnd_keys=[(frm1_key, frm2_key)])
rcts_atm_keys = list(map(atom_keys, rct_gras))
prds_atm_keys = list(map(atom_keys, prd_gras))
if frm2_key not in prds_atm_keys[1]:
prds_atm_keys = list(reversed(prds_atm_keys))
# Create the reaction object
rxns.append(
Reaction(
rxn_cls=par.ReactionClass.Typ.ELIMINATION,
forw_tsg=forw_tsg,
back_tsg=back_tsg,
rcts_keys=rcts_atm_keys,
prds_keys=prds_atm_keys,
))
if viable_only:
rxns = filter_viable_reactions(rxns)
return ts_unique(rxns)
def hydrogen_migrations(rct_gras, prd_gras):
""" find hydrogen migrations consistent with these reactants and products
:param rct_gras: reactant graphs (must have non-overlapping keys)
:param prd_gras: product graphs (must have non-overlapping keys)
Hydrogen migrations are identified by adding a hydrogen to an unsaturated
site of the reactant and adding a hydrogen to an unsaturated site of the
product and seeing if they match up. If so, we have a hydrogen migration
between these two sites.
"""
_assert_is_valid_reagent_graph_list(rct_gras)
_assert_is_valid_reagent_graph_list(prd_gras)
rxns = []
if len(rct_gras) == 1 and len(prd_gras) == 1:
rct_gra, = rct_gras
prd_gra, = prd_gras
# Find keys for reactant graph
rct_h_key = max(atom_keys(rct_gra)) + 1
rct_rad_keys = unsaturated_atom_keys(rct_gra)
# Find keys for product graph
prd_h_key = max(atom_keys(prd_gra)) + 1
prd_rad_keys = unsaturated_atom_keys(prd_gra)
for rct_rad_key, prd_rad_key in (itertools.product(
rct_rad_keys, prd_rad_keys)):
# Add hydrogens to each radical site and see if the result matches
rct_h_gra = add_bonded_atom(rct_gra,
'H',
rct_rad_key,
bnd_atm_key=rct_h_key)
prd_h_gra = add_bonded_atom(prd_gra,
'H',
prd_rad_key,
bnd_atm_key=prd_h_key)
iso_dct = isomorphism(rct_h_gra, prd_h_gra)
if iso_dct:
inv_dct = dict(map(reversed, iso_dct.items()))
rct_don_key = inv_dct[prd_rad_key]
prd_don_key = iso_dct[rct_rad_key]
# Check equivalent donor atoms for other possible TSs
rct_don_keys = equivalent_atoms(rct_h_gra, rct_don_key)
prd_don_keys = equivalent_atoms(prd_h_gra, prd_don_key)
for rct_don_key, prd_don_key in (itertools.product(
rct_don_keys, prd_don_keys)):
rct_hyd_key = atom_neighbor_atom_key(rct_gra,
rct_don_key,
symbs_first=('H', ),
symbs_last=())
prd_hyd_key = atom_neighbor_atom_key(prd_gra,
prd_don_key,
symbs_first=('H', ),
symbs_last=())
forw_tsg = ts.graph(rct_gra,
frm_bnd_keys=[(rct_rad_key,
rct_hyd_key)],
brk_bnd_keys=[(rct_don_key,
rct_hyd_key)])
back_tsg = ts.graph(prd_gra,
frm_bnd_keys=[(prd_rad_key,
prd_hyd_key)],
brk_bnd_keys=[(prd_don_key,
prd_hyd_key)])
if isomorphism(forw_tsg, ts.reverse(back_tsg)):
rxns.append(
Reaction(
rxn_cls=par.ReactionClass.HYDROGEN_MIGRATION,
forw_tsg=forw_tsg,
back_tsg=back_tsg,
rcts_keys=[atom_keys(rct_gra)],
prds_keys=[atom_keys(prd_gra)],
))
return ts_unique(rxns)
def eliminations(rct_gras, prd_gras):
""" find eliminations consistent with these reactants and products
:param rct_gras: reactant graphs (must have non-overlapping keys)
:param prd_gras: product graphs (must have non-overlapping keys)
Eliminations are identified by forming a bond between an attacking heavy
atom and another atom not initially bonded to it, forming a ring. The bond
adjacent to the attacked atom is then broken, along with a second bond in
the ring, downstream of the attacking heavy atom, away from the attacked
atom.
"""
_assert_is_valid_reagent_graph_list(rct_gras)
_assert_is_valid_reagent_graph_list(prd_gras)
rxns = []
if len(rct_gras) == 1 and len(prd_gras) == 2:
rct_gra, = rct_gras
prds_gra = union_from_sequence(prd_gras)
ngb_keys_dct = atoms_neighbor_atom_keys(rct_gra)
frm1_keys = atom_keys(rct_gra, excl_syms=('H', ))
frm2_keys = atom_keys(rct_gra)
bnd_keys = bond_keys(rct_gra)
frm_bnd_keys = [
(frm1_key, frm2_key)
for frm1_key, frm2_key in itertools.product(frm1_keys, frm2_keys)
if frm1_key != frm2_key
and not frozenset({frm1_key, frm2_key}) in bnd_keys
]
for frm1_key, frm2_key in frm_bnd_keys:
# Bond the radical atom to the hydrogen atom
gra_ = add_bonds(rct_gra, [(frm2_key, frm1_key)])
# Get keys to the ring formed by this extra bond
rng_keys = next((ks for ks in rings_atom_keys(gra_)
if frm2_key in ks and frm1_key in ks), None)
# Eliminations (as far as I can tell) only happen through TSs with
# 3- or 4-membered rings
if rng_keys is not None and len(rng_keys) < 5:
frm1_ngb_key, = ngb_keys_dct[frm1_key] & set(rng_keys)
frm2_ngb_key, = ngb_keys_dct[frm2_key] & set(rng_keys)
# Break the bonds on either side of the newly formed bond
gra_ = remove_bonds(gra_, [(frm1_key, frm1_ngb_key)])
gra_ = remove_bonds(gra_, [(frm2_key, frm2_ngb_key)])
inv_dct = isomorphism(gra_, prds_gra)
if inv_dct:
f_frm_bnd_key = (frm1_key, frm2_key)
f_brk_bnd_key1 = (frm1_key, frm1_ngb_key)
f_brk_bnd_key2 = (frm2_key, frm2_ngb_key)
inv_ = inv_dct.__getitem__
b_frm_bnd_key1 = tuple(map(inv_, f_brk_bnd_key1))
b_frm_bnd_key2 = tuple(map(inv_, f_brk_bnd_key2))
b_brk_bnd_key = tuple(map(inv_, f_frm_bnd_key))
forw_tsg = ts.graph(
rct_gra,
frm_bnd_keys=[f_frm_bnd_key],
brk_bnd_keys=[f_brk_bnd_key1, f_brk_bnd_key2])
back_tsg = ts.graph(
prds_gra,
frm_bnd_keys=[b_frm_bnd_key1, b_frm_bnd_key2],
brk_bnd_keys=[b_brk_bnd_key])
rcts_atm_keys = list(map(atom_keys, rct_gras))
prds_atm_keys = list(map(atom_keys, prd_gras))
if inv_dct[frm2_key] not in prds_atm_keys[1]:
prds_atm_keys = list(reversed(prds_atm_keys))
# Create the reaction object
rxns.append(
Reaction(
rxn_cls=par.ReactionClass.ELIMINATION,
forw_tsg=forw_tsg,
back_tsg=back_tsg,
rcts_keys=rcts_atm_keys,
prds_keys=prds_atm_keys,
))
return ts_unique(rxns)
def substitutions(rct_gras, prd_gras):
""" find substitutions consistent with these reactants and products
:param rct_gras: reactant graphs (must have non-overlapping keys)
:param prd_gras: product graphs (must have non-overlapping keys)
Substitutions are identified by breaking one bond in the reactants and one
bond from the products and checking for isomorphism.
Currently it assumes that one of the reactants has a radical site that
can attack the other reactants, forming a bond and breaking another.
From the perspective of breaking and forming breaking bonds, substitutions
are equivalent with hydrogen abstractions. Hence, we remove all cases where
the forming bond involves a hydrogen atom off the reactant in which a bond
is breaking.
"""
assert_is_valid_reagent_graph_list(rct_gras)
assert_is_valid_reagent_graph_list(prd_gras)
rxns = []
if len(rct_gras) == 2 and len(prd_gras) == 2:
rct_gra = union_from_sequence(rct_gras)
prd_gra = union_from_sequence(prd_gras)
# Loop over both orders of reactants: A+B and B+A
for rgra1, rgra2 in itertools.permutations(rct_gras):
bnd_keys = bond_keys(rgra1)
atom_symb_dct = automol.graph.atom_symbols(rgra1)
rad_keys = unsaturated_atom_keys(rgra2)
# Break all possible bonds in total reactant
for bnd_key, rad_key in itertools.product(bnd_keys, rad_keys):
gra = remove_bonds(rct_gra, [bnd_key])
# Form all possible bonds between rad site and non-H atoms
frm_keys = ()
for key in bnd_key:
frm_symb = atom_symb_dct[key]
if frm_symb != 'H':
frm_keys += (key, )
for frm_key in frm_keys:
gra = add_bonds(gra, [(frm_key, rad_key)])
inv_dct = isomorphism(gra, prd_gra)
if inv_dct:
brk_key2, = bnd_key - {frm_key}
f_frm_bnd_key = (frm_key, rad_key)
f_brk_bnd_key = (frm_key, brk_key2)
b_frm_bnd_key = (inv_dct[frm_key], inv_dct[brk_key2])
b_brk_bnd_key = (inv_dct[frm_key], inv_dct[rad_key])
forw_tsg = ts.graph(rct_gra,
frm_bnd_keys=[f_frm_bnd_key],
brk_bnd_keys=[f_brk_bnd_key])
back_tsg = ts.graph(prd_gra,
frm_bnd_keys=[b_frm_bnd_key],
brk_bnd_keys=[b_brk_bnd_key])
rcts_atm_keys = [atom_keys(rgra1), atom_keys(rgra2)]
prds_atm_keys = list(map(atom_keys, prd_gras))
if inv_dct[rad_key] not in prds_atm_keys[0]:
prds_atm_keys = list(reversed(prds_atm_keys))
# Create the reaction object
rxns.append(
Reaction(
rxn_cls=ReactionClass.Typ.SUBSTITUTION,
forw_tsg=forw_tsg,
back_tsg=back_tsg,
rcts_keys=rcts_atm_keys,
prds_keys=prds_atm_keys,
))
return ts_unique(rxns)
def eliminations(rct_gras, prd_gras):
""" find eliminations consistent with these reactants and products
:param rct_gras: reactant graphs (must have non-overlapping keys)
:param prd_gras: product graphs (must have non-overlapping keys)
Eliminations are identified by forming a bond between an attacking heavy
atom and another atom not initially bonded to it, forming a ring. The bond
adjacent to the attacked atom is then broken, along with a second bond in
the ring, downstream of the attacking heavy atom, away from the attacked
atom.
"""
def _identify(frm1_keys, frm2_keys, bnd_keys):
""" Try and identify elmination from some set of keys
"""
_rxns = []
frm_bnd_keys = [
(frm1_key, frm2_key)
for frm1_key, frm2_key in itertools.product(frm1_keys, frm2_keys)
if frm1_key != frm2_key
and not frozenset({frm1_key, frm2_key}) in bnd_keys
]
for frm1_key, frm2_key in frm_bnd_keys:
prds_gra_ = add_bonds(rct_gra, [(frm2_key, frm1_key)])
# Get keys of all bonds in the ring formed by this extra bond
rng_bnd_keys = next((ks for ks in rings_bond_keys(prds_gra_)
if frozenset({frm1_key, frm2_key}) in ks),
None)
if rng_bnd_keys is not None:
# Elims break two bonds of the ring formed by the forming bond
# Loop over all ring bond-pairs, break bonds, see if prods form
# Ensure to preclude the forming-bond from this set
brk_bnds = tuple(
bond for bond in itertools.combinations(rng_bnd_keys, 2)
if frozenset({frm1_key, frm2_key}) not in bond)
for brk_bnd_1, brk_bnd_2 in brk_bnds:
prds_gra_2_ = prds_gra_
prds_gra_2_ = remove_bonds(prds_gra_2_, [brk_bnd_1])
prds_gra_2_ = remove_bonds(prds_gra_2_, [brk_bnd_2])
inv_dct = isomorphism(prds_gra_2_, prds_gra)
if inv_dct:
f_frm_bnd_key = (frm1_key, frm2_key)
inv_ = inv_dct.__getitem__
b_frm_bnd_key1 = tuple(map(inv_, brk_bnd_1))
b_frm_bnd_key2 = tuple(map(inv_, brk_bnd_2))
b_brk_bnd_key = tuple(map(inv_, f_frm_bnd_key))
forw_tsg = ts.graph(
rct_gra,
frm_bnd_keys=[f_frm_bnd_key],
brk_bnd_keys=[brk_bnd_1, brk_bnd_2])
back_tsg = ts.graph(
prds_gra,
frm_bnd_keys=[b_frm_bnd_key1, b_frm_bnd_key2],
brk_bnd_keys=[b_brk_bnd_key])
rcts_atm_keys = list(map(atom_keys, rct_gras))
prds_atm_keys = list(map(atom_keys, prd_gras))
if inv_dct[frm1_key] not in prds_atm_keys[1]:
prds_atm_keys = list(reversed(prds_atm_keys))
assert inv_dct[frm1_key] in prds_atm_keys[1]
assert inv_dct[frm2_key] in prds_atm_keys[1]
# Create the reaction object
_rxns.append(
Reaction(
rxn_cls=ReactionClass.Typ.ELIMINATION,
forw_tsg=forw_tsg,
back_tsg=back_tsg,
rcts_keys=rcts_atm_keys,
prds_keys=prds_atm_keys,
))
return _rxns
assert_is_valid_reagent_graph_list(rct_gras)
assert_is_valid_reagent_graph_list(prd_gras)
rxns = []
if len(rct_gras) == 1 and len(prd_gras) == 2:
rct_gra, = rct_gras
prds_gra = union_from_sequence(prd_gras)
# ngb_keys_dct = atoms_neighbor_atom_keys(rct_gra)
# Generate keys all bonds and 1/2 the forming bond
frm1_keys = atom_keys(rct_gra)
bnd_keys = bond_keys(rct_gra)
frm2_keys = unsaturated_atom_keys(rct_gra)
rct_symbs = atom_symbols(rct_gra)
frm2_keys_o = frozenset(key for key in frm2_keys
if rct_symbs[key] == 'O')
rxns.extend(_identify(frm1_keys, frm2_keys_o, bnd_keys))
# OLD WAY. More IDs but more mistakes
# To make the function general, try to ID reaction
# with different types of keys for the attacking atom
# (1) unsaturated atom sites
# frm2_keys = unsaturated_atom_keys(rct_gra)
# rxns.extend(_identify(frm1_keys, frm2_keys, bnd_keys))
# if not rxns:
# # (2) remaining saturated atom sites
# frm2_keys = atom_keys(rct_gra, excl_syms=('H',)) - frm2_keys
# rxns.extend(_identify(frm1_keys, frm2_keys, bnd_keys))
# # if not rxns: # Ignoring H2 formation for now for speed
# # # (3) H atoms
# # frm1_keys = atom_keys(rct_gra, sym='H')
# # rxns.extend(_identify(frm1_keys, frm2_keys, bnd_keys))
return ts_unique(rxns)
def hydrogen_migrations(rct_gras, prd_gras):
""" find hydrogen migrations consistent with these reactants and products
:param rct_gras: reactant graphs (must have non-overlapping keys)
:param prd_gras: product graphs (must have non-overlapping keys)
Hydrogen migrations are identified by adding a hydrogen to an unsaturated
site of the reactant and adding a hydrogen to an unsaturated site of the
product and seeing if they match up. If so, we have a hydrogen migration
between these two sites.
"""
_assert_is_valid_reagent_graph_list(rct_gras)
_assert_is_valid_reagent_graph_list(prd_gras)
rxns = []
if len(rct_gras) == 1 and len(prd_gras) == 1:
gra1, = rct_gras
gra2, = prd_gras
# Get the keys for the reactant graph
h_atm_key1 = max(atom_keys(gra1)) + 1
atm_keys1 = unsaturated_atom_keys(gra1)
# Generate reactions for all isomorphic graphs of products
gra2_lst = (gra2, ) + isomorphic_radical_graphs(gra2)
# gra2_lst = (gra2,)
for _gra2 in gra2_lst:
# Find keys for product graph
h_atm_key2 = max(atom_keys(_gra2)) + 1
atm_keys2 = unsaturated_atom_keys(_gra2)
# Run identifier
for atm_key1, atm_key2 in itertools.product(atm_keys1, atm_keys2):
gra1_h = add_atom_explicit_hydrogen_keys(
gra1, {atm_key1: [h_atm_key1]})
gra2_h = add_atom_explicit_hydrogen_keys(
_gra2, {atm_key2: [h_atm_key2]})
iso_dct = full_isomorphism(gra1_h, gra2_h)
if iso_dct:
inv_dct = dict(map(reversed, iso_dct.items()))
f_frm_bnd_key = (atm_key1, inv_dct[h_atm_key2])
f_brk_bnd_key = (inv_dct[atm_key2], inv_dct[h_atm_key2])
b_frm_bnd_key = (atm_key2, iso_dct[h_atm_key1])
b_brk_bnd_key = (iso_dct[atm_key1], iso_dct[h_atm_key1])
forw_tsg = ts.graph(gra1,
frm_bnd_keys=[f_frm_bnd_key],
brk_bnd_keys=[f_brk_bnd_key])
back_tsg = ts.graph(_gra2,
frm_bnd_keys=[b_frm_bnd_key],
brk_bnd_keys=[b_brk_bnd_key])
# Create the reaction object
rxns.append(
Reaction(
rxn_cls=par.ReactionClass.HYDROGEN_MIGRATION,
forw_tsg=forw_tsg,
back_tsg=back_tsg,
rcts_keys=[atom_keys(gra1)],
prds_keys=[atom_keys(_gra2)],
))
return tuple(rxns)
def eliminations(rct_gras, prd_gras):
""" find eliminations consistent with these reactants and products
:param rct_gras: reactant graphs (must have non-overlapping keys)
:param prd_gras: product graphs (must have non-overlapping keys)
Eliminations are identified by forming a bond between an attacking heavy
atom and another atom not initially bonded to it, forming a ring. The bond
adjacent to the attacked atom is then broken, along with a second bond in
the ring, downstream of the attacking heavy atom, away from the attacked
atom.
"""
_assert_is_valid_reagent_graph_list(rct_gras)
_assert_is_valid_reagent_graph_list(prd_gras)
rxns = []
if len(rct_gras) == 1 and len(prd_gras) == 2:
rgra, = rct_gras
pgra = union_from_sequence(prd_gras)
rngb_keys = atoms_sorted_neighbor_atom_keys(rgra)
frm1_keys = atom_keys(rgra, excl_syms=('H', ))
frm2_keys = atom_keys(rgra)
bnd_keys = bond_keys(rgra)
frm_bnd_keys = [
(frm1_key, frm2_key)
for frm1_key, frm2_key in itertools.product(frm1_keys, frm2_keys)
if frm1_key != frm2_key
and not frozenset({frm1_key, frm2_key}) in bnd_keys
]
for frm1_key, frm2_key in frm_bnd_keys:
# Bond the radical atom to the hydrogen atom
rgra_ = add_bonds(rgra, [(frm2_key, frm1_key)])
# Get keys to the ring formed by this extra bond
rng_keys = next((ks for ks in rings_atom_keys(rgra_)
if frm2_key in ks and frm1_key in ks), None)
if rng_keys is not None:
for nfrm2_key in rngb_keys[frm2_key]:
# Break the bond between the attacked atom and its neighbor
rgra_ = remove_bonds(rgra_, [(frm2_key, nfrm2_key)])
# Sort the ring keys so that they start with the radical
# atom and end with the hydrogen atom
keys = cycle_ring_atom_key_to_front(rng_keys,
frm1_key,
end_key=frm2_key)
# Break one ring bond at a time, starting from the rind,
# and see what we get
for brk_key1, brk_key2 in mit.windowed(keys[:-1], 2):
gra = remove_bonds(rgra_, [(brk_key1, brk_key2)])
inv_dct = full_isomorphism(gra, pgra)
if inv_dct:
f_frm_bnd_key = (frm2_key, frm1_key)
f_brk_bnd_key1 = (frm2_key, nfrm2_key)
f_brk_bnd_key2 = (brk_key1, brk_key2)
b_frm_bnd_key1 = (inv_dct[frm2_key],
inv_dct[nfrm2_key])
b_frm_bnd_key2 = (inv_dct[brk_key1],
inv_dct[brk_key2])
b_brk_bnd_key = (inv_dct[frm2_key],
inv_dct[frm1_key])
forw_tsg = ts.graph(
rgra,
frm_bnd_keys=[f_frm_bnd_key],
brk_bnd_keys=[f_brk_bnd_key1, f_brk_bnd_key2])
back_tsg = ts.graph(
pgra,
frm_bnd_keys=[b_frm_bnd_key1, b_frm_bnd_key2],
brk_bnd_keys=[b_brk_bnd_key])
rcts_atm_keys = list(map(atom_keys, rct_gras))
prds_atm_keys = list(map(atom_keys, prd_gras))
if inv_dct[frm2_key] not in prds_atm_keys[1]:
prds_atm_keys = list(reversed(prds_atm_keys))
# Create the reaction object
rxns.append(
Reaction(
rxn_cls=par.ReactionClass.ELIMINATION,
forw_tsg=forw_tsg,
back_tsg=back_tsg,
rcts_keys=rcts_atm_keys,
prds_keys=prds_atm_keys,
))
return tuple(rxns)
