def _set_bond_stereo_from_geometry(gra, bnd_keys, geo, geo_idx_dct):
assert gra == explicit(gra)
bnd_pars = [
_bond_stereo_parity_from_geometry(gra, bnd_key, geo, geo_idx_dct)
for bnd_key in bnd_keys
]
gra = set_bond_stereo_parities(gra, dict(zip(bnd_keys, bnd_pars)))
return gra
def _set_atom_stereo_from_geometry(gra, atm_keys, geo, geo_idx_dct):
assert gra == explicit(gra)
atm_pars = [
_atom_stereo_parity_from_geometry(gra, atm_key, geo, geo_idx_dct)
for atm_key in atm_keys
]
gra = set_atom_stereo_parities(gra, dict(zip(atm_keys, atm_pars)))
return gra
def _connected_heuristic_geometry(gra):
""" stereo-specific coordinates for a connected molecular geometry
"""
assert gra == explicit(gra)
atm_keys = sorted(atom_keys(gra))
zma, zma_key_dct = connected_heuristic_zmatrix(gra)
geo = automol.zmatrix.geometry(zma)
idxs = dict_.values_by_key(zma_key_dct, atm_keys)
geo = automol.geom.from_subset(geo, idxs)
geo_idx_dct = {atm_key: idx for idx, atm_key in enumerate(atm_keys)}
return geo, geo_idx_dct
def stereogenic_bond_keys(gra):
""" (unassigned) stereogenic bonds in this graph
"""
gra = without_bond_orders(gra)
gra = explicit(gra) # for simplicity, add the explicit hydrogens back in
bnd_keys = dict_.keys_by_value(
resonance_dominant_bond_orders(gra), lambda x: 2 in x)
# make sure both ends are sp^2 (excludes cumulenes)
atm_hyb_dct = resonance_dominant_atom_hybridizations(gra)
sp2_atm_keys = dict_.keys_by_value(atm_hyb_dct, lambda x: x == 2)
bnd_keys = frozenset({bnd_key for bnd_key in bnd_keys
if bnd_key <= sp2_atm_keys})
bnd_keys -= bond_stereo_keys(gra)
bnd_keys -= functools.reduce( # remove double bonds in small rings
frozenset.union,
filter(lambda x: len(x) < 8, rings_bond_keys(gra)), frozenset())
atm_ngb_keys_dct = atom_neighbor_keys(gra)
def _is_stereogenic(bnd_key):
atm1_key, atm2_key = bnd_key
def _is_symmetric_on_bond(atm_key, atm_ngb_key):
atm_ngb_keys = list(atm_ngb_keys_dct[atm_key] - {atm_ngb_key})
if not atm_ngb_keys: # C=:O:
ret = True
elif len(atm_ngb_keys) == 1: # C=N:-X
ret = False
else:
assert len(atm_ngb_keys) == 2 # C=C(-X)-Y
ret = (stereo_priority_vector(gra, atm_key, atm_ngb_keys[0]) ==
stereo_priority_vector(gra, atm_key, atm_ngb_keys[1]))
return ret
return not (_is_symmetric_on_bond(atm1_key, atm2_key) or
_is_symmetric_on_bond(atm2_key, atm1_key))
ste_gen_bnd_keys = frozenset(filter(_is_stereogenic, bnd_keys))
return ste_gen_bnd_keys
def stereogenic_atom_keys(gra):
""" (unassigned) stereogenic atoms in this graph
"""
gra = without_bond_orders(gra)
gra = explicit(gra) # for simplicity, add the explicit hydrogens back in
atm_keys = dict_.keys_by_value(atom_bond_valences(gra), lambda x: x == 4)
atm_keys -= atom_stereo_keys(gra)
atm_ngb_keys_dct = atom_neighbor_keys(gra)
def _is_stereogenic(atm_key):
atm_ngb_keys = list(atm_ngb_keys_dct[atm_key])
pri_vecs = [stereo_priority_vector(gra, atm_key, atm_ngb_key)
for atm_ngb_key in atm_ngb_keys]
return not any(pv1 == pv2
for pv1, pv2 in itertools.combinations(pri_vecs, r=2))
ste_gen_atm_keys = frozenset(filter(_is_stereogenic, atm_keys))
return ste_gen_atm_keys
def heuristic_geometry(gra):
""" stereo-specific coordinates for a molecular geometry
(need not be connected)
"""
assert gra == explicit(gra)
gra_iter = iter(connected_components(gra))
gra_ = next(gra_iter)
geo_, geo_idx_dct_ = _connected_heuristic_geometry(gra_)
geo = geo_
geo_idx_dct = geo_idx_dct_
for gra_ in gra_iter:
geo_, geo_idx_dct_ = _connected_heuristic_geometry(gra_)
natms = automol.geom.count(geo)
geo_idx_dct_ = dict_.transform_values(geo_idx_dct_, (natms).__add__)
geo_idx_dct.update(geo_idx_dct_)
geo = automol.geom.join(geo, geo_)
return geo, geo_idx_dct
def _stereo_corrected_geometry(sgr, geo, geo_idx_dct):
""" correct the stereo parities of a geometry
(works iterately to handle cases of higher-order stereo)
"""
assert sgr == explicit(sgr)
gra = without_stereo_parities(sgr)
if has_stereo(sgr):
full_atm_ste_par_dct = atom_stereo_parities(sgr)
full_bnd_ste_par_dct = bond_stereo_parities(sgr)
atm_keys = set()
bnd_keys = set()
last_gra = None
while last_gra != gra:
last_gra = gra
atm_keys.update(stereogenic_atom_keys(gra))
bnd_keys.update(stereogenic_bond_keys(gra))
atm_ste_par_dct = {
atm_key: full_atm_ste_par_dct[atm_key]
for atm_key in atm_keys
}
bnd_ste_par_dct = {
bnd_key: full_bnd_ste_par_dct[bnd_key]
for bnd_key in bnd_keys
}
geo, gra = _atom_stereo_corrected_geometry(gra, atm_ste_par_dct,
geo, geo_idx_dct)
geo, gra = _bond_stereo_corrected_geometry(gra, bnd_ste_par_dct,
geo, geo_idx_dct)
return geo
def connected_heuristic_zmatrix(gra):
""" stereo-specific coordinates for a connected molecular graph
(currently unable to handle rings -- fix that)
"""
assert gra == explicit(gra)
atm_sym_dct = atom_symbols(gra)
# this will contain triplets of adjacent atoms from which to continue
# filling out the z-matrix, after it has been started
triplets = []
# 1. start the z-matrix and set the lists of triplets
rng_atm_keys_lst = rings_atom_keys(gra)
if not rng_atm_keys_lst:
# find the first heavy atom in the longest chain (if there isn't one,
# we are dealing with atomic or molecular hydrogen, which will be
# captured by the last two cases)
chain = longest_chain(gra)
if atm_sym_dct[chain[0]] != 'H':
chain = list(reversed(chain))
if len(chain) > 1:
atm_key = chain[1]
else:
atm_key = chain[0]
# determine the z-matrix of the starting atom and its neighbors
zma, zma_key_dct, dummy_atm_key, gra = _start_zmatrix_from_atom(
gra, atm_key)
# since this is the first heavy atom in the longest chain, we only need
# to follow one branch from this atom to complete the z-matrix; this
# will be the branch extending toward the next heavy atom in the chai
if len(chain) > 3:
atm1_key, atm2_key, atm3_key = chain[:3]
# if we inserted a dummy atom on the starting geometry, we should
# use that as atom 1 in the triplet, rather than
if dummy_atm_key is not None:
atm1_key = dummy_atm_key
triplets = [(atm1_key, atm2_key, atm3_key)]
elif len(rng_atm_keys_lst) == 1:
rng_atm_keys, = rng_atm_keys_lst
zma, zma_key_dct = _start_zmatrix_from_ring(gra, rng_atm_keys)
triplets += list(mit.windowed(rng_atm_keys[-2:] + rng_atm_keys, 3))
else:
# currently, multiple rings are not implemented
raise NotImplementedError
# 2. complete the z-matrix by looping over triplets
for atm1_key, atm2_key, atm3_key in triplets:
zma, zma_key_dct, gra = _complete_zmatrix_for_branch(
gra, atm1_key, atm2_key, atm3_key, zma, zma_key_dct)
# 3. convert to Cartesian geometry for stereo correction
geo = automol.zmatrix.geometry(zma)
geo_idx_dct = zma_key_dct
geo = _stereo_corrected_geometry(gra, geo, geo_idx_dct)
# 4. convert back to z-matrix, keeping the original z-matrix structure
vma = automol.zmatrix.var_(zma)
zma = automol.zmatrix.from_geometry(vma, geo)
return zma, zma_key_dct
4: ('C', 1, True),
5: ('O', 1, None),
6: ('O', 0, None),
7: ('O', 0, None)
}, {
frozenset({3, 4}): (1, None),
frozenset({2, 6}): (1, None),
frozenset({0, 2}): (1, None),
frozenset({3, 6}): (1, None),
frozenset({2, 4}): (1, None),
frozenset({1, 3}): (1, None),
frozenset({5, 7}): (1, None),
frozenset({4, 7}): (1, None)
})
SGR = explicit(GRA)
GEO, GEO_IDX_DCT = heuristic_geometry(SGR)
# GEO = (
# ('C', (-4.3870588134, -1.233231672517, 0.143749726309016)),
# ('C', (3.430304171771, -2.162836645393, 0.06129774977456508)),
# ('C', (-2.228277354885, 0.1940343942502, -1.0788747507898575)),
# ('C', (1.642516616594, -0.2666792157215, -1.1217150146524835)),
# ('C', (-0.128929182575, 1.167922277159, 0.6891116967734786)),
# ('O', (2.44416862025, 4.3088944278, 2.2811617948010614)),
# ('O', (-0.4839163664245, -1.530973627393, -2.314392676447687)),
# ('O', (0.1546996690877, 3.88389710099, 0.8186943317590577)),
# ('H', (-3.69835752847, -2.73865502820, 1.385388988006255)),
# ('H', (-5.55611791656, 0.04579783881378, 1.2732325623382277)),
# ('H', (-5.59710374244, -2.09637223693, -1.2950602194746856)),
# ('H', (2.43356758828, -3.46358524781, 1.3249230383597128)),
def _are_all_explicit(gras):
return all(gra == explicit(gra) for gra in gras)