def combine_ligands_and_hierarchy(pdb_hierarchy, ligands, log=None):
from iotbx.pdb import hierarchy
if (log is None) : log = null_out()
chain_id_counts = {}
model = pdb_hierarchy.models()[0]
for i_lig, ligand in enumerate(ligands):
xyz_mean = ligand.atoms().extract_xyz().mean()
best_chain = None
min_dist = sys.maxsize
for chain in model.chains():
last_resseq = chain.residue_groups()[-1].resseq_as_int()
if ((not chain.id in chain_id_counts) or
(chain_id_counts[chain.id] < last_resseq)):
chain_id_counts[chain.id] = last_resseq
if (not chain.is_protein()) : continue
chain_xyz_mean = chain.atoms().extract_xyz().mean()
dist = xyz_distance(chain_xyz_mean, xyz_mean)
if (dist < min_dist):
min_dist = dist
best_chain = chain
best_chain_id = " "
if (best_chain is not None):
best_chain_id = best_chain.id
new_chain = hierarchy.chain(id=best_chain_id)
new_rg = hierarchy.residue_group()
new_resseq = 1
if (best_chain_id in chain_id_counts):
new_resseq = chain_id_counts[best_chain_id] + 1
print(" ligand %d: chain='%s' resseq=%s" % (i_lig+1,
best_chain_id, new_resseq), file=log)
new_rg.resseq = new_resseq
new_rg.append_atom_group(ligand)
new_chain.append_residue_group(new_rg)
model.append_chain(new_chain)
chain_id_counts[best_chain_id] = new_resseq
def combine_ligands_and_hierarchy (pdb_hierarchy, ligands, log=None) :
from iotbx.pdb import hierarchy
if (log is None) : log = null_out()
chain_id_counts = {}
model = pdb_hierarchy.models()[0]
for i_lig, ligand in enumerate(ligands) :
xyz_mean = ligand.atoms().extract_xyz().mean()
best_chain = None
min_dist = sys.maxint
for chain in model.chains() :
last_resseq = chain.residue_groups()[-1].resseq_as_int()
if ((not chain.id in chain_id_counts) or
(chain_id_counts[chain.id] < last_resseq)) :
chain_id_counts[chain.id] = last_resseq
if (not chain.is_protein()) : continue
chain_xyz_mean = chain.atoms().extract_xyz().mean()
dist = xyz_distance(chain_xyz_mean, xyz_mean)
if (dist < min_dist) :
min_dist = dist
best_chain = chain
best_chain_id = " "
if (best_chain is not None) :
best_chain_id = best_chain.id
new_chain = hierarchy.chain(id=best_chain_id)
new_rg = hierarchy.residue_group()
new_resseq = 1
if (best_chain_id in chain_id_counts) :
new_resseq = chain_id_counts[best_chain_id] + 1
print >> log, " ligand %d: chain='%s' resseq=%s" % (i_lig+1,
best_chain_id, new_resseq)
new_rg.resseq = new_resseq
new_rg.append_atom_group(ligand)
new_chain.append_residue_group(new_rg)
model.append_chain(new_chain)
chain_id_counts[best_chain_id] = new_resseq
def __init__(self, cif_block):
crystal_symmetry_builder.__init__(self, cif_block)
self.hierarchy = hierarchy.root()
# These items are mandatory for the _atom_site loop, all others are optional
type_symbol = self._wrap_loop_if_needed(cif_block,
"_atom_site.type_symbol")
atom_labels = self._wrap_loop_if_needed(cif_block,
"_atom_site.auth_atom_id")
if atom_labels is None:
atom_labels = self._wrap_loop_if_needed(
cif_block, "_atom_site.label_atom_id"
) # corresponds to chem comp atom name
alt_id = self._wrap_loop_if_needed(
cif_block, "_atom_site.label_alt_id") # alternate conformer id
label_asym_id = self._wrap_loop_if_needed(
cif_block, "_atom_site.label_asym_id") # chain id
auth_asym_id = self._wrap_loop_if_needed(cif_block,
"_atom_site.auth_asym_id")
if label_asym_id is None: label_asym_id = auth_asym_id
if auth_asym_id is None: auth_asym_id = label_asym_id
comp_id = self._wrap_loop_if_needed(cif_block,
"_atom_site.auth_comp_id")
if comp_id is None:
comp_id = self._wrap_loop_if_needed(
cif_block, "_atom_site.label_comp_id") # residue name
entity_id = self._wrap_loop_if_needed(cif_block,
"_atom_site.label_entity_id")
seq_id = self._wrap_loop_if_needed(cif_block, "_atom_site.auth_seq_id")
if seq_id is None:
seq_id = self._wrap_loop_if_needed(
cif_block, "_atom_site.label_seq_id") # residue number
assert [atom_labels, alt_id, auth_asym_id, comp_id, entity_id,
seq_id].count(None) == 0, "someting is not present"
assert type_symbol is not None
atom_site_fp = cif_block.get('_atom_site.phenix_scat_dispersion_real')
atom_site_fdp = cif_block.get('_atom_site.phenix_scat_dispersion_imag')
pdb_ins_code = cif_block.get(
"_atom_site.pdbx_PDB_ins_code") # insertion code
model_ids = cif_block.get("_atom_site.pdbx_PDB_model_num")
atom_site_id = cif_block.get("_atom_site.id")
# only permitted values are ATOM or HETATM
group_PDB = cif_block.get("_atom_site.group_PDB")
# TODO: read esds
B_iso_or_equiv = flex.double(
self._wrap_loop_if_needed(cif_block, "_atom_site.B_iso_or_equiv"))
cart_x = flex.double(
self._wrap_loop_if_needed(cif_block, "_atom_site.Cartn_x"))
cart_y = flex.double(
self._wrap_loop_if_needed(cif_block, "_atom_site.Cartn_y"))
cart_z = flex.double(
self._wrap_loop_if_needed(cif_block, "_atom_site.Cartn_z"))
occu = flex.double(
self._wrap_loop_if_needed(cif_block, "_atom_site.occupancy"))
formal_charge = self._wrap_loop_if_needed(
cif_block, "_atom_site.pdbx_formal_charge")
# anisotropic b-factors
# TODO: read esds
anisotrop_id = self._wrap_loop_if_needed(cif_block,
"_atom_site_anisotrop.id")
adps = None
if anisotrop_id is not None:
u_ij = [
self._wrap_loop_if_needed(
cif_block,
"_atom_site_anisotrop.U[%s][%s]" % (ij[0], ij[1]))
for ij in ("11", "22", "33", "12", "13", "23")
]
assert u_ij.count(None) in (0, 6)
if u_ij.count(None) == 0:
adps = u_ij
else:
assert u_ij.count(None) == 6
b_ij = [
self._wrap_loop_if_needed(
cif_block,
"_atom_site_anisotrop.B[%s][%s]" % (ij[0], ij[1]))
for ij in ("11", "22", "33", "12", "13", "23")
]
assert b_ij.count(None) in (0, 6)
if b_ij.count(None) == 0:
adps = adptbx.b_as_u(b_ij)
assert not (u_ij.count(None) and b_ij.count(None)
) # illegal for both to be present
if adps is not None:
try:
adps = [flex.double(adp) for adp in adps]
except ValueError as e:
raise CifBuilderError("Error interpreting ADPs: " + str(e))
adps = flex.sym_mat3_double(*adps)
py_adps = {}
if anisotrop_id is not None and adps is not None:
for an_id, adp in zip(list(anisotrop_id), list(adps)):
py_adps[an_id] = adp
current_model_id = None
current_label_asym_id = None
current_auth_asym_id = None
current_residue_id = None
current_ins_code = None
for i_atom in range(atom_labels.size()):
# model(s)
last_model_id = current_model_id
current_model_id = model_ids[i_atom]
assert current_model_id is not None
if current_model_id != last_model_id:
model = hierarchy.model(id=current_model_id)
self.hierarchy.append_model(model)
# chain(s)
last_label_asym_id = current_label_asym_id
current_label_asym_id = label_asym_id[i_atom]
assert current_label_asym_id is not None
last_auth_asym_id = current_auth_asym_id
current_auth_asym_id = auth_asym_id[i_atom]
assert current_auth_asym_id not in [".", "?", " "], "mmCIF file contains " + \
"record with empty auth_asym_id, which is wrong."
assert current_label_asym_id is not None
if (current_auth_asym_id != last_auth_asym_id
or current_model_id != last_model_id):
chain = hierarchy.chain(id=current_auth_asym_id)
model.append_chain(chain)
else:
assert current_auth_asym_id == last_auth_asym_id
# residue_group(s)
# defined by residue id and insertion code
last_residue_id = current_residue_id
current_residue_id = seq_id[i_atom]
assert current_residue_id is not None
last_ins_code = current_ins_code
if pdb_ins_code is not None:
current_ins_code = pdb_ins_code[i_atom]
if current_ins_code in ("?", ".", None): current_ins_code = " "
if (current_residue_id != last_residue_id
or current_ins_code != last_ins_code
or current_auth_asym_id != last_auth_asym_id
or current_model_id != last_model_id):
try:
resseq = hy36encode(width=4, value=int(current_residue_id))
except ValueError as e:
resseq = current_residue_id
assert len(resseq) == 4
residue_group = hierarchy.residue_group(resseq=resseq,
icode=current_ins_code)
chain.append_residue_group(residue_group)
atom_groups = OrderedDict() # reset atom_groups cache
# atom_group(s)
# defined by resname and altloc id
current_altloc = alt_id[i_atom]
if current_altloc == "." or current_altloc == "?":
current_altloc = "" # Main chain atoms
current_resname = comp_id[i_atom]
if (current_altloc, current_resname) not in atom_groups:
atom_group = hierarchy.atom_group(altloc=current_altloc,
resname=current_resname)
atom_groups[(current_altloc, current_resname)] = atom_group
if current_altloc == "":
residue_group.insert_atom_group(0, atom_group)
else:
residue_group.append_atom_group(atom_group)
else:
atom_group = atom_groups[(current_altloc, current_resname)]
# atom(s)
atom = hierarchy.atom()
atom_group.append_atom(atom)
atom.set_element(type_symbol[i_atom])
atom.set_name(
format_pdb_atom_name(atom_labels[i_atom], type_symbol[i_atom]))
atom.set_xyz(new_xyz=(cart_x[i_atom], cart_y[i_atom],
cart_z[i_atom]))
atom.set_b(B_iso_or_equiv[i_atom])
atom.set_occ(occu[i_atom])
# hy36encode should go once the pdb.hierarchy has been
# modified to no longer store fixed-width strings
atom.set_serial(
hy36encode(width=5, value=int(atom_site_id[i_atom])))
# some code relies on an empty segid being 4 spaces
atom.set_segid(" ")
if group_PDB is not None and group_PDB[i_atom] == "HETATM":
atom.hetero = True
if formal_charge is not None:
charge = formal_charge[i_atom]
if charge not in ("?", "."):
if charge.endswith("-") or charge.startswith("-"):
sign = "-"
else:
sign = "+"
charge = charge.strip(" -+")
charge = int(charge)
if charge == 0: sign = ""
atom.set_charge("%i%s" % (charge, sign))
if atom_site_fp is not None:
fp = atom_site_fp[i_atom]
if fp not in ("?", "."):
atom.set_fp(new_fp=float(fp))
if atom_site_fdp is not None:
fdp = atom_site_fdp[i_atom]
if fdp not in ("?", "."):
atom.set_fdp(new_fdp=float(fdp))
if anisotrop_id is not None and adps is not None:
py_u_ij = py_adps.get(atom.serial.strip(), None)
if py_u_ij is not None:
atom.set_uij(py_u_ij)
if len(self.hierarchy.models()) == 1:
# for compatibility with single-model PDB files
self.hierarchy.models()[0].id = ""
def __init__(self,
pdb_hierarchy,
fmodel,
ncs_operators,
ligands,
params,
log=None):
if (log is None) : log = sys.stdout
if (params is None):
params = master_phil().fetch().extract()
adopt_init_args(self, locals())
self.xray_structure = fmodel.xray_structure.deep_copy_scatterers()
xrs_ncs = fmodel.xray_structure.deep_copy_scatterers()
from iotbx.pdb import hierarchy
self.setup_maps()
best_cc = 0
best_k = -1
best_ligand = None
other_ligands = []
print("Identifying reference ligand...", file=log)
def show_map_stats(prefix, stats):
print(" %s: CC = %5.3f mean = %6.2f" % (prefix, stats.cc,
stats.map_mean), file=log)
for k, ligand in enumerate(ligands):
atoms = ligand.atoms()
start = self.get_sites_cc(atoms)
show_map_stats("Ligand %d" % (k+1), start)
if (start.cc > best_cc) and (start.cc > params.min_cc_reference):
best_ligand = ligand
best_k = k
best_cc = best_cc
if (best_ligand is None):
raise Sorry("No ligand with acceptable CC (>%.2f) found." %
params.min_cc_reference)
best_atoms = best_ligand.atoms()
for k, ligand in enumerate(ligands):
if (ligand is not best_ligand):
other_ligands.append(ligand)
print("Copy #%d was the best, using that as reference" % (best_k+1), file=log)
print("", file=log)
sites_ref = best_ligand.atoms().extract_xyz()
min_dist = sys.maxsize
best_group = None
shifts = ncs_operators.get_ncs_groups_shifts(
self.xray_structure.sites_cart(),
sites_ref
)
for i, s in enumerate(shifts):
dxyz = xyz_distance(s[0], (0,0,0))
if (dxyz < min_dist):
best_group = i
min_dist = dxyz
array_of_str_selections = ncs_operators.get_array_of_str_selections()[0]
if (best_group is not None):
print("This appears to be bound to the selection \"%s\"" % \
array_of_str_selections[best_group], file=log)
if best_group==0: pass
else:
print('best_group',best_group)
assert 0
# always have the first ligand in the master
self.new_ligands = []
for j, operator in enumerate(ncs_operators[0].copies):
new_ligand = best_ligand.detached_copy()
atoms = new_ligand.atoms()
sites_new = operator.r.elems * sites_ref + operator.t.elems
sites_mean = sites_new.mean()
for other in other_ligands :
sites_other_mean = other.atoms().extract_xyz().mean()
dxyz = xyz_distance(sites_other_mean, sites_new.mean())
if (dxyz < params.min_dist_center):
print(" operator %d specifies an existing ligand" % (j+1), file=log)
break
else :
atoms.set_xyz(sites_new)
stats_new = self.get_sites_cc(best_atoms, sites_new)
show_map_stats("NCS op. %2d" % (j+1), stats_new)
if (params.write_sampled_pdbs):
lig_rg = hierarchy.residue_group()
lig_rg.resseq = j+1
lig_rg.append_atom_group(new_ligand)
f = open("ncs_ligand_%d.pdb" % (j+1), "w")
for atom in new_ligand.atoms():
f.write(atom.format_atom_record()+"\n")
f.close()
# XXX ideally, given multiple high-quality ligand placements, we should
# probably try sampling NCS operations for all of these and pick the
# best new CC, rather than assuming that the best starting ligand will
# superpose best on the density.
if (stats_new.cc > params.min_cc):
print(" operator %d has acceptable CC (%.3f)" % (j+1,
stats_new.cc), file=log)
self.new_ligands.append(new_ligand)
def __init__ (self,
pdb_hierarchy,
fmodel,
ncs_operators,
ligands,
params,
log=None) :
if (log is None) : log = sys.stdout
if (params is None) :
params = master_phil().fetch().extract()
adopt_init_args(self, locals())
self.xray_structure = fmodel.xray_structure.deep_copy_scatterers()
xrs_ncs = fmodel.xray_structure.deep_copy_scatterers()
from iotbx.pdb import hierarchy
self.setup_maps()
best_cc = 0
best_k = -1
best_ligand = None
other_ligands = []
print >> log, "Identifying reference ligand..."
def show_map_stats (prefix, stats) :
print >> log, " %s: CC = %5.3f mean = %6.2f" % (prefix, stats.cc,
stats.map_mean)
for k, ligand in enumerate(ligands) :
atoms = ligand.atoms()
start = self.get_sites_cc(atoms)
show_map_stats("Ligand %d" % (k+1), start)
if (start.cc > best_cc) and (start.cc > params.min_cc_reference) :
best_ligand = ligand
best_k = k
best_cc = best_cc
if (best_ligand is None) :
raise Sorry("No ligand with acceptable CC (>%.2f) found." %
params.min_cc_reference)
best_atoms = best_ligand.atoms()
for k, ligand in enumerate(ligands) :
if (ligand is not best_ligand) :
other_ligands.append(ligand)
print >> log, "Copy #%d was the best, using that as reference" % (best_k+1)
print >> log, ""
sites_ref = best_ligand.atoms().extract_xyz()
min_dist = sys.maxint
best_group = None
for op_group in ncs_operators :
dxyz = op_group.distance_from_center(sites_ref)
if (dxyz < min_dist) :
best_group = op_group
min_dist = dxyz
if (best_group is not None) :
print >> log, "This appears to be bound to the selection \"%s\"" % \
best_group.selection_string
self.new_ligands = []
for j, operator in enumerate(best_group.operators) :
new_ligand = best_ligand.detached_copy()
atoms = new_ligand.atoms()
sites_new = operator.r.elems * sites_ref + operator.t.elems
sites_mean = sites_new.mean()
for other in other_ligands :
sites_other_mean = other.atoms().extract_xyz().mean()
dxyz = xyz_distance(sites_other_mean, sites_new.mean())
if (dxyz < params.min_dist_center) :
print >> log, " operator %d specifies an existing ligand" % (j+1)
break
else :
atoms.set_xyz(sites_new)
stats_new = self.get_sites_cc(best_atoms, sites_new)
show_map_stats("NCS op. %2d" % (j+1), stats_new)
if (params.write_sampled_pdbs) :
lig_rg = hierarchy.residue_group()
lig_rg.resseq = j+1
lig_rg.append_atom_group(new_ligand)
f = open("ncs_ligand_%d.pdb" % (j+1), "w")
for atom in new_ligand.atoms() :
f.write(atom.format_atom_record()+"\n")
f.close()
# XXX ideally, given multiple high-quality ligand placements, we should
# probably try sampling NCS operations for all of these and pick the
# best new CC, rather than assuming that the best starting ligand will
# superpose best on the density.
if (stats_new.cc > params.min_cc) :
print >> log, " operator %d has acceptable CC (%.3f)" % (j+1,
stats_new.cc)
self.new_ligands.append(new_ligand)