def __init__(self,
fixed_ref_atoms,
moving_h,
moving_ref_atoms_iseqs,
max_number_of_iterations=500,
needed_rmsd=0.1):
"""
fixed_ref_atoms - list of 3 atom objects, actually, only xyz's are needed
moving_ref_atoms_iseqs - list of 3 indeces matching atoms in
moving_h.atoms()[<here!>].
moving_h - hierarchy to make closure. Atom positions in it will be changed!
"""
assert len(fixed_ref_atoms) == 3
assert len(moving_ref_atoms_iseqs) == 3
assert moving_h is not None
assert moving_h.atoms_size() > 10 # arbitrary
# adopt_init_args(self, locals())
self.moving_h = moving_h
self.fixed_ref_atoms = fixed_ref_atoms
self.moving_ref_atoms_iseqs = moving_ref_atoms_iseqs
self.max_number_of_iterations = max_number_of_iterations
self.needed_rmsd = needed_rmsd
self.set_modify_angle_procedure(self._modify_angle)
self.r = rama_eval()
# self.states = mmtbx.utils.states(pdb_hierarchy=moving_h)
self.convergence_diff = 1e-5
# will be bool, True if converged before max_number_of_iterations reached
self.early_exit = None
self.resulting_rmsd = None
def get_all_starting_conformations(moving_h, change_radius, cutoff=50, log=null_out()):
variants = []
r = rama_eval()
phi_psi_atoms = utils.get_phi_psi_atoms(moving_h)
n_rama = len(phi_psi_atoms)
change_angles = range((n_rama)//2-change_radius, (n_rama)//2+change_radius+1)
# print " change_angles", change_angles
for i, (phi_psi_pair, rama_key) in enumerate(phi_psi_atoms):
if i in change_angles or (utils.rama_evaluate(phi_psi_pair, r, rama_key) == ramalyze.RAMALYZE_OUTLIER):
if utils.rama_evaluate(phi_psi_pair, r, rama_key) == ramalyze.RAMALYZE_OUTLIER:
vs = get_sampled_rama_favored_angles(rama_key, r)
else:
vs = ramalyze.get_favored_regions(rama_key)
variants.append(vs)
# variants.append(ramalyze.get_favored_regions(rama_key))
else:
variants.append([(None, None)])
print >> log, "variants", variants
all_angles_combination = list(itertools.product(*variants))
result = []
i = 0
n_added = 0
n_all_combination = len(all_angles_combination)
i_max = min(cutoff, n_all_combination)
while n_added < i_max:
comb = all_angles_combination[i]
if is_not_none_combination(comb):
result.append(set_rama_angles(moving_h, list(comb)))
print >> log, "Model %d, angles:" % i, comb
n_added += 1
i += 1
# STOP()
return result
def get_starting_conformations(moving_h, cutoff=50, log=null_out()):
"""
modify only ramachandran outliers.
"""
variants = []
r = rama_eval()
phi_psi_atoms = utils.get_phi_psi_atoms(moving_h)
for phi_psi_pair, rama_key in phi_psi_atoms:
if (utils.rama_evaluate(phi_psi_pair, r, rama_key) == ramalyze.RAMALYZE_OUTLIER):
vs = get_sampled_rama_favored_angles(rama_key, r)
# print len(vs)
# print vs
# STOP()
variants.append(vs)
# variants.append(ramalyze.get_favored_regions(rama_key))
else:
variants.append([(None, None)])
result = []
print >> log, "variants", variants
if variants.count([(None, None)]) == len(variants):
print "Nothing to CCD"
return result
all_angles_combination = list(itertools.product(*variants))
i = 0
n_added = 0
n_all_combination = len(all_angles_combination)
i_max = min(cutoff, n_all_combination)
while n_added < i_max:
comb = all_angles_combination[i]
if is_not_none_combination(comb):
result.append(set_rama_angles(moving_h, list(comb)))
print >> log, "Model %d, angles:" % i, comb
n_added += 1
i += 1
return result
def __init__(self, fixed_ref_atoms, moving_h, moving_ref_atoms_iseqs,
max_number_of_iterations=500, needed_rmsd=0.1):
"""
fixed_ref_atoms - list of 3 atom objects, actually, only xyz's are needed
moving_ref_atoms_iseqs - list of 3 indeces matching atoms in
moving_h.atoms()[<here!>].
moving_h - hierarchy to make closure. Atom positions in it will be changed!
"""
assert len(fixed_ref_atoms) == 3
assert len(moving_ref_atoms_iseqs) == 3
assert moving_h is not None
assert moving_h.atoms_size() > 10 # arbitrary
# adopt_init_args(self, locals())
self.moving_h = moving_h
self.fixed_ref_atoms = fixed_ref_atoms
self.moving_ref_atoms_iseqs = moving_ref_atoms_iseqs
self.max_number_of_iterations = max_number_of_iterations
self.needed_rmsd = needed_rmsd
self.set_modify_angle_procedure(self._modify_angle)
self.r = rama_eval()
# self.states = mmtbx.utils.states(pdb_hierarchy=moving_h)
self.convergence_diff = 1e-5
# will be bool, True if converged before max_number_of_iterations reached
self.early_exit = None
self.resulting_rmsd = None
def get_all_starting_conformations(moving_h,
change_radius,
n_outliers,
direction_forward=True,
cutoff=50,
change_all=True,
log=null_out(),
check_omega=False):
if log is None:
log = StringIO()
variants = []
result = []
r = rama_eval()
phi_psi_atoms = utils.get_phi_psi_atoms(moving_h, omega=True)
# print "N residue groups in h", [x.resseq for x in moving_h.residue_groups()]
if len(phi_psi_atoms) == 0:
print >> log, "Strange input to starting conformations!!!"
return result
n_rama = len(phi_psi_atoms)
# print "n_rama", n_rama
change_angles = [None]
if change_all:
change_angles = range(
(n_rama) // 2 - change_radius - n_outliers // 2,
(n_rama) // 2 + change_radius + 1 + n_outliers // 2)
# if change_angles[0] < 0:
# change_angles = range(change_angles[-1]-change_angles[0])
has_twisted = False
if check_omega:
omegas = [x[2] for x in phi_psi_atoms]
for o in omegas:
if o is not None and abs(abs(o) - 180) > 30:
has_twisted = True
print >> log, "n_outliers", n_outliers
for i, (phi_psi_pair, rama_key, omega) in enumerate(phi_psi_atoms):
angle_is_outlier = utils.rama_evaluate(
phi_psi_pair, r, rama_key) == ramalyze.RAMALYZE_OUTLIER
twisted = omega is not None and ((abs(abs(omega) - 180) > 30)
and check_omega)
print >> log, "in cycle, N, outlier?, change?, twisted?", i, angle_is_outlier, i in change_angles, twisted
if angle_is_outlier and n_outliers < 3:
vs = get_sampled_rama_favored_angles(rama_key, r)
elif (i in change_angles) or angle_is_outlier or has_twisted:
# vs = get_sampled_rama_favored_angles(rama_key, r)
vs = ramalyze.get_favored_regions(rama_key)
else:
vs = [(None, None)]
variants.append(vs)
print >> log, "variants", variants
all_angles_combination = list(itertools.product(*variants))
# filter none combinations
# print "len(all_angles_combination)", len(all_angles_combination)
all_angles_combination_f = []
for comb in all_angles_combination:
if is_not_none_combination(comb):
all_angles_combination_f.append(comb)
print >> log, "len(all_angles_combination_f)", len(
all_angles_combination_f)
return all_angles_combination_f
def list_rama_outliers_h(hierarchy, r=None, include_allowed=False):
if r is None:
r = rama_eval()
phi_psi_atoms = get_phi_psi_atoms(hierarchy)
outp = list_rama_outliers(phi_psi_atoms,
r,
include_allowed=include_allowed)
return outp
def extract_proxies(self, hierarchy):
self.hierarchy = hierarchy
selected_h = hierarchy.select(self.bool_atom_selection)
n_seq = flex.max(selected_h.atoms().extract_i_seq())
# Drop all previous proxies
self._oldfield_proxies = ext.shared_phi_psi_proxy()
self._emsley_proxies = ext.shared_phi_psi_proxy()
# it would be great to save rama_eval, but the fact that this is called in
# pdb_interpretation, not in mmtbx.model makes it impossible
if self.need_filtering:
self.rama_eval = rama_eval()
for three in generate_protein_threes(hierarchy=selected_h,
geometry=None):
rc = three.get_phi_psi_atoms()
if rc is None: continue
rama_key = three.get_ramalyze_key()
if self.need_filtering:
angles = three.get_phi_psi_angles()
rama_score = self.rama_eval.get_score(rama_key, angles[0],
angles[1])
r_evaluation = self.rama_eval.evaluate_score(
rama_key, rama_score)
phi_atoms, psi_atoms = rc
i_seqs = [atom.i_seq for atom in phi_atoms] + [psi_atoms[-1].i_seq]
resnames = three.get_resnames()
r_name = resnames[1]
assert rama_key in range(6)
text_rama_key = ramalyze.res_types[rama_key]
assert text_rama_key in [
"general", "glycine", "cis-proline", "trans-proline",
"pre-proline", "isoleucine or valine"
]
proxy = ext.phi_psi_proxy(residue_name=r_name,
residue_type=text_rama_key,
i_seqs=i_seqs)
# pick where to put...
if self.params.rama_potential == "oldfield":
if self.need_filtering:
if r_evaluation == ramalyze.RAMALYZE_FAVORED:
self.append_oldfield_proxies(proxy, n_seq)
elif r_evaluation == ramalyze.RAMALYZE_ALLOWED and self.params.restrain_rama_allowed:
self.append_oldfield_proxies(proxy, n_seq)
elif r_evaluation == ramalyze.RAMALYZE_OUTLIER and self.params.restrain_rama_outliers:
self.append_oldfield_proxies(proxy, n_seq)
elif self.params.restrain_allowed_outliers_with_emsley:
self.append_emsley_proxies(proxy, n_seq)
else:
self.append_oldfield_proxies(proxy, n_seq)
else: # self.params.rama_potential == "emsley":
self.append_emsley_proxies(proxy, n_seq)
print("", file=self.log)
print(" %d Ramachandran restraints generated." %
(self.get_n_proxies()),
file=self.log)
print(" %d Oldfield and %d Emsley." %
(self.get_n_oldfield_proxies(), self.get_n_emsley_proxies()),
file=self.log)
def get_sampled_rama_favored_angles(rama_key, r=None, step=20):
if r is None:
r = rama_eval()
result = []
for i in range(-180, 180, step):
for j in range(-180, 180, step):
score = r.evaluate_angles(ramalyze.res_types[rama_key], i,j)
r_ev = ramalyze.ramalyze.evalScore(ramalyze.res_types[rama_key], score)
if r_ev == ramalyze.RAMALYZE_FAVORED:
result.append((i,j))
return result
def get_sampled_rama_favored_angles(rama_key, r=None, step=20):
if r is None:
r = rama_eval()
result = []
for i in range(-180, 180, step):
for j in range(-180, 180, step):
score = r.evaluate_angles(ramalyze.res_types[rama_key], i,j)
r_ev = ramalyze.ramalyze.evalScore(ramalyze.res_types[rama_key], score)
if r_ev == ramalyze.RAMALYZE_FAVORED:
result.append((i,j))
return result
def rama_outliers_selection(hierarchy, r=None, margin=1):
if r is None:
r = rama_eval()
out_sel = []
phi_psi_atoms = get_phi_psi_atoms(hierarchy)
for phi_psi_pair, rama_key in phi_psi_atoms:
rama_score = get_rama_score(phi_psi_pair, r, rama_key)
if rama_evaluate(phi_psi_pair, r, rama_key) == ramalyze.RAMALYZE_OUTLIER:
out_sel.append(pair_selection(phi_psi_pair, margin))
out_sel_txt = " or ".join(out_sel)
return out_sel_txt
def rama_outliers_selection(hierarchy, r=None, margin=1):
if r is None:
r = rama_eval()
out_sel = []
phi_psi_atoms = get_phi_psi_atoms(hierarchy)
for phi_psi_pair, rama_key in phi_psi_atoms:
rama_score = get_rama_score(phi_psi_pair, r, rama_key)
if rama_evaluate(phi_psi_pair, r,
rama_key) == ramalyze.RAMALYZE_OUTLIER:
out_sel.append(pair_selection(phi_psi_pair, margin))
out_sel_txt = " or ".join(out_sel)
return out_sel_txt
def __init__(self,
pdb_h,
mediocre_hbond_cutoff=3.0,
bad_hbond_cutoff=3.5,
rama_eval_manager=None):
self.pdb_h = pdb_h
self.bad_hbond_cutoff = bad_hbond_cutoff
self.mediocre_hbond_cutoff = mediocre_hbond_cutoff
self.asc = self.pdb_h.atom_selection_cache()
self.atoms = pdb_h.atoms()
self.r = rama_eval_manager
if self.r is None:
self.r = rama_eval()
def rama_score_selection(hierarchy, r=None, score="outlier", margin=1):
assert score in ["outlier", "allowed"]
test = ramalyze.RAMALYZE_OUTLIER
if score == "allowed":
test = ramalyze.RAMALYZE_ALLOWED
if r is None:
r = rama_eval()
out_sel = []
phi_psi_atoms = get_phi_psi_atoms(hierarchy)
for phi_psi_pair, rama_key in phi_psi_atoms:
rama_score = get_rama_score(phi_psi_pair, r, rama_key)
if rama_evaluate(phi_psi_pair, r, rama_key) == test:
out_sel.append(pair_selection(phi_psi_pair, margin))
out_sel_txt = " or ".join(out_sel)
return out_sel_txt
def __init__(self,
pdb_input,
cif_objects=None,
params=None,
log=sys.stdout,
verbose=True):
t_0 = time()
self.pdb_input = pdb_input
self.cif_objects = cif_objects
self.params = params
self.log = log
self.verbose = verbose
self.rmsd_from_start = None
self.init_model_statistics = None
self.after_ss_idealization = None
self.after_loop_idealization = None
self.after_rotamer_fixing = None
self.final_model_statistics = None
self.reference_map = None
self.whole_grm = None
self.master_grm = None
self.working_grm = None
self.mon_lib_srv = None
self.ener_lib = None
self.rotamer_manager = None
self.rama_manager = rama_eval()
self.original_hierarchy = None # original pdb_h, without any processing
self.original_boxed_hierarchy = None # original and boxed (if needed)
self.whole_pdb_h = None # boxed with processing (AC trimming, H trimming,...)
self.master_pdb_h = None # master copy in case of NCS
self.working_pdb_h = None # one to use for fixing (master_pdb_h or working_pdb_h)
# various checks, shifts, trims
self.cs = self.pdb_input.crystal_symmetry()
# check self.cs (copy-paste from secondary_sturcure_restraints)
corrupted_cs = False
if self.cs is not None:
if [self.cs.unit_cell(), self.cs.space_group()].count(None) > 0:
corrupted_cs = True
self.cs = None
elif self.cs.unit_cell().volume() < 10:
corrupted_cs = True
self.cs = None
self.original_hierarchy = self.pdb_input.construct_hierarchy()
# couple checks if pdb_h is ok
o_c = self.original_hierarchy.overall_counts()
o_c.raise_duplicate_atom_labels_if_necessary()
o_c.raise_residue_groups_with_multiple_resnames_using_same_altloc_if_necessary()
o_c.raise_chains_with_mix_of_proper_and_improper_alt_conf_if_necessary()
o_c.raise_improper_alt_conf_if_necessary()
if len(self.original_hierarchy.models()) > 1:
raise Sorry("Multi model files are not supported")
ca_only_present = False
for c in self.original_hierarchy.only_model().chains():
if c.is_ca_only():
ca_only_present = True
if ca_only_present:
raise Sorry("Don't support models with chains containing only CA atoms.")
self.original_boxed_hierarchy = self.original_hierarchy.deep_copy()
self.original_boxed_hierarchy.reset_atom_i_seqs()
self.shift_vector = None
if self.cs is None:
if corrupted_cs:
print >> self.log, "Symmetry information is corrupted, "
else:
print >> self.log, "Symmetry information was not found, "
print >> self.log, "putting molecule in P1 box."
self.log.flush()
from cctbx import uctbx
atoms = self.original_boxed_hierarchy.atoms()
box = uctbx.non_crystallographic_unit_cell_with_the_sites_in_its_center(
sites_cart=atoms.extract_xyz(),
buffer_layer=3)
atoms.set_xyz(new_xyz=box.sites_cart)
self.cs = box.crystal_symmetry()
self.shift_vector = box.shift_vector
# self.original_boxed_hierarchy.write_pdb_file(file_name="original_boxed_h.pdb")
if self.shift_vector is not None:
write_whole_pdb_file(
file_name="%s_boxed.pdb" % self.params.output_prefix,
pdb_hierarchy=self.original_boxed_hierarchy,
crystal_symmetry=self.cs,
ss_annotation=self.pdb_input.extract_secondary_structure())
asc = self.original_boxed_hierarchy.atom_selection_cache()
if self.params.trim_alternative_conformations:
sel = asc.selection("altloc ' '")
self.whole_pdb_h = self.original_boxed_hierarchy.select(sel).deep_copy()
print >> self.log, "Atoms in original/working model: %d/%d" % (
self.original_boxed_hierarchy.atoms_size(), self.whole_pdb_h.atoms_size())
else:
self.whole_pdb_h = self.original_boxed_hierarchy.deep_copy()
# self.whole_pdb_h.reset_atom_i_seqs()
# Trimming hydrogens
# Many intermediate variables are needed due to strange behavior of
# selections described in
# iotbx/pdb/tst_hierarchy.py:exercise_selection_and_deep_copy()
asc2 = self.whole_pdb_h.atom_selection_cache()
h_sel = asc2.selection("not (element H or element D)")
temp_h = self.whole_pdb_h.select(h_sel)
self.whole_pdb_h = temp_h.deep_copy()
self.whole_pdb_h.reset_atom_i_seqs()
self.init_model_statistics = geometry_no_grm(
pdb_hierarchy=iotbx.pdb.input(
source_info=None,
lines=self.whole_pdb_h.as_pdb_string()).construct_hierarchy(),
molprobity_scores=True)
self.time_for_init = time()-t_0
def list_rama_outliers_h(hierarchy, r=None):
if r is None:
r = rama_eval()
phi_psi_atoms = get_phi_psi_atoms(hierarchy)
outp = list_rama_outliers(phi_psi_atoms, r)
return outp
def __init__(self,
pdb_hierarchy,
params=None,
log=sys.stdout,
proxies=None,
tables=None,
initialize=True):
assert pdb_hierarchy is not None
assert not pdb_hierarchy.atoms().extract_i_seq().all_eq(0), ""+\
"Probably all atoms have i_seq = 0 which is wrong"
if params is None:
# print ('init, params is None')
w_params = master_phil.fetch().extract()
w_params = w_params.ramachandran_plot_restraints
elif hasattr(params, 'enabled'):
# print ("init, hasattr(params, 'enabled')")
# New params
w_params = params
elif hasattr(params, 'ramachandran_plot_restraints'):
# print ("init, hasattr(params, 'ramachandran_plot_restraints'")
# print ("init, ", type(params), type(params.ramachandran_plot_restraints), params.ramachandran_plot_restraints)
w_params = params.ramachandran_plot_restraints
else:
# print ("init, else")
w_params = master_phil.fetch().extract()
w_params = w_params.ramachandran_plot_restraints
# old params, make transfer
w_params.selection = params.rama_selection
# oldfield
w_params.enabled = True
w_params.oldfield.weight = \
params.oldfield.weight if (params.oldfield.weight is None or params.oldfield.weight > 0) else 0
w_params.oldfield.weight_scale = \
1/(params.oldfield.esd**2) * params.oldfield.weight_scale
w_params.oldfield.distance_weight_min = 2.0
w_params.oldfield.distance_weight_max = params.oldfield.dist_weight_max
# emsley
w_params.emsley.weight = params.rama_weight
w_params.emsley.scale_allowed = params.scale_allowed
# strategy
if params.rama_potential == 'oldfield':
pass
elif params.rama_potential == 'emsley':
w_params.favored = 'emsley'
w_params.allowed = 'emsley'
w_params.outlier = 'emsley'
if params.restrain_rama_outliers:
w_params.outlier = params.rama_potential
else:
w_params.outlier = None
if params.restrain_rama_allowed:
w_params.allowed = params.rama_potential
else:
w_params.allowed = None
if params.restrain_allowed_outliers_with_emsley:
if not params.restrain_rama_allowed:
w_params.allowed = 'emsley'
if not params.restrain_rama_outliers:
w_params.outlier = 'emsley'
self.params = w_params
self.rama_eval = rama_eval()
self.hierarchy = pdb_hierarchy # only for def select()
self.log = log
self._oldfield_proxies = ext.shared_phi_psi_proxy()
self._emsley_proxies = ext.shared_phi_psi_proxy()
self._emsley8k_proxies = ext.shared_phi_psi_proxy()
self._phi_psi_2_proxies = ext.shared_phi_psi_proxy()
self._oldfield_tables = None
self._emsley_tables = None
self._emsley8k_tables = None
self._phi_psi_2_tables = None
if proxies is not None:
self._oldfield_proxies, \
self._emsley_proxies, \
self._emsley8k_proxies, \
self._phi_psi_2_proxies = proxies
if tables is not None:
self._oldfield_tables, \
self._emsley_tables, \
self._emsley8k_tables, \
self._phi_psi_2_tables = tables
self.initialize = initialize
# bad hack to keep emsley potential in working(?) condition after
# changing from rama500 to rama8000
self.new_to_old_conversion = {
"general": "ala",
"glycine": "gly",
"cis-proline": "pro",
"trans-proline": "pro",
"pre-proline": "prepro",
"isoleucine or valine": "ala"
}
bool_atom_selection = self._determine_bool_atom_selection(
pdb_hierarchy)
fao = [self.params.favored, self.params.allowed, self.params.outlier]
if initialize:
if 'oldfield' in fao:
self._oldfield_tables = ramachandran_plot_data(
plot_cutoff=self.params.oldfield.plot_cutoff)
if 'emsley' in fao:
self._emsley_tables = load_tables()
#
### THIS IS CRUEL. REMOVE ONCE favored/allowed/outlier are made multiple!
#
if 'emsley8k' in fao or self.params.inject_emsley8k_into_oldfield_favored:
self._emsley8k_tables = load_emsley8k_tables()
if 'phi_psi_2' in fao:
self._phi_psi_2_tables = load_phi_psi_2_tables()
# get proxies
self.extract_proxies(pdb_hierarchy)
if 'oldfield' in fao:
self.target_phi_psi = self.update_phi_psi_targets_on_init(
hierarchy=pdb_hierarchy)
self.initialize = False
def extract_proxies(self, hierarchy):
def _get_motifs():
from phenix.programs.phi_psi_2 import results_manager as pp2
pp2_manager = pp2(model=None, log=self.log)
phi_psi_2_motifs = pp2_manager.get_overall_motif_count_and_output(
None,
self.hierarchy,
return_rama_restraints=True,
)
return phi_psi_2_motifs
phi_psi_2_motifs = None
favored = ramalyze.RAMALYZE_FAVORED
allowed = ramalyze.RAMALYZE_ALLOWED
outlier = ramalyze.RAMALYZE_OUTLIER
self.hierarchy = hierarchy
bool_atom_selection = self._determine_bool_atom_selection(hierarchy)
selected_h = hierarchy.select(bool_atom_selection)
n_seq = flex.max(selected_h.atoms().extract_i_seq())
# Drop all previous proxies
self._oldfield_proxies = ext.shared_phi_psi_proxy()
self._emsley_proxies = ext.shared_phi_psi_proxy()
self._emsley8k_proxies = ext.shared_phi_psi_proxy()
self._phi_psi_2_proxies = ext.shared_phi_psi_proxy()
# it would be great to save rama_eval, but the fact that this is called in
# pdb_interpretation, not in mmtbx.model makes it impossible
self.rama_eval = rama_eval()
outl = []
for three in generate_protein_threes(hierarchy=selected_h,
geometry=None):
rc = three.get_phi_psi_atoms()
if rc is None: continue
rama_key = three.get_ramalyze_key()
angles = three.get_phi_psi_angles()
rama_score = self.rama_eval.get_score(rama_key, angles[0],
angles[1])
r_eval = self.rama_eval.evaluate_score(rama_key, rama_score)
phi_atoms, psi_atoms = rc
i_seqs = [atom.i_seq for atom in phi_atoms] + [psi_atoms[-1].i_seq]
resnames = three.get_resnames()
r_name = resnames[1]
assert rama_key in range(6)
text_rama_key = ramalyze.res_types[rama_key]
assert text_rama_key in [
"general", "glycine", "cis-proline", "trans-proline",
"pre-proline", "isoleucine or valine"
]
# pick where to put...
ev_match_dict = {
favored: self.params.favored,
allowed: self.params.allowed,
outlier: self.params.outlier
}
r_type = ev_match_dict[r_eval]
if r_type == 'oldfield':
proxy = ext.phi_psi_proxy(residue_type=text_rama_key,
i_seqs=i_seqs,
weight=1) # XXX Not used in oldfield
self.append_oldfield_proxies(proxy, n_seq)
### THIS IS CRUEL. REMOVE ONCE favored/allowed/outlier are made multiple!
if (self.params.inject_emsley8k_into_oldfield_favored):
proxy = ext.phi_psi_proxy(residue_type=text_rama_key,
i_seqs=i_seqs,
weight=5)
self.append_emsley8k_proxies(proxy, n_seq)
###
elif r_type == 'emsley':
weight = self.params.emsley.weight
proxy = ext.phi_psi_proxy(residue_type=text_rama_key,
i_seqs=i_seqs,
weight=weight)
self.append_emsley_proxies(proxy, n_seq)
elif r_type == 'emsley8k':
if (r_eval is favored):
weight = self.params.emsley8k.weight_favored
elif (r_eval is allowed):
weight = self.params.emsley8k.weight_allowed
elif (r_eval is outlier):
weight = self.params.emsley8k.weight_outlier
else:
raise RuntimeError("Rama eveluation failed.")
proxy = ext.phi_psi_proxy(residue_type=text_rama_key,
i_seqs=i_seqs,
weight=weight)
self.append_emsley8k_proxies(proxy, n_seq)
elif r_type == 'phi_psi_2':
from phenix.pdb_tools.phi_psi_2_data import get_phi_psi_key_for_rama_proxy
if phi_psi_2_motifs is None: phi_psi_2_motifs = _get_motifs()
if (r_eval is favored):
strategy = self.params.phi_psi_2.favored_strategy
elif (r_eval is allowed):
strategy = self.params.phi_psi_2.allowed_strategy
elif (r_eval is outlier):
strategy = self.params.phi_psi_2.outlier_strategy
else:
raise RuntimeError("Rama eveluation failed.")
if strategy == 'closest':
strategy += '_%0.1f_%0.1f' % tuple(
three.get_phi_psi_angles())
pp2_key = get_phi_psi_key_for_rama_proxy(
phi_psi_2_motifs,
three,
strategy=strategy,
)
if pp2_key is None: continue
weight = 1
proxy = ext.phi_psi_proxy(residue_type=pp2_key,
i_seqs=i_seqs,
weight=weight)
outl.append([proxy.residue_type, three])
self.append_phi_psi_2_proxies(proxy, n_seq)
elif (r_type is None):
pass
else:
raise RuntimeError("Not an option: %s" % str(r_type))
print("", file=self.log)
print(" %d Ramachandran restraints generated." %
(self.get_n_proxies()),
file=self.log)
print(" %d Oldfield, %d Emsley, %d emsley8k and %d Phi/Psi/2." %
(self.get_n_oldfield_proxies(), self.get_n_emsley_proxies(),
self.get_n_emsley8k_proxies(), self.get_n_phi_psi_2_proxies()),
file=self.log)
if outl:
print(' Rama restraints by Phi/Psi/2')
for pp2, three in outl:
print(' %s : %s' % (three[1].id_str(), pp2.split('|')[0]),
file=self.log)
def __init__(self): self.rama_eval = rama_eval()
def __init__(self):
self.rama_eval = rama_eval()
def list_rama_outliers_h(hierarchy, r=None):
if r is None:
r = rama_eval()
phi_psi_atoms = get_phi_psi_atoms(hierarchy)
outp = list_rama_outliers(phi_psi_atoms, r)
return outp
def __init__(self,
pdb_hierarchy,
params=None,
secondary_structure_annotation=None,
reference_map=None,
crystal_symmetry=None,
grm=None,
rama_manager=None,
rotamer_manager=None,
log=null_out(),
verbose=False,
tried_rama_angles={},
tried_final_rama_angles={},
n_run=0):
if len(pdb_hierarchy.models()) > 1:
raise Sorry("Multi-model files are not supported")
self.original_pdb_h = pdb_hierarchy
self.secondary_structure_annotation=secondary_structure_annotation
asc = pdb_hierarchy.atom_selection_cache()
self.xrs = pdb_hierarchy.extract_xray_structure(crystal_symmetry=crystal_symmetry)
self.reference_map = reference_map
self.resulting_pdb_h = pdb_hierarchy.deep_copy()
self.resulting_pdb_h.reset_atom_i_seqs()
self.params = self.process_params(params)
self.log = log
self.verbose = verbose
self.grm = grm
self.r = rama_manager
self.ideal_res_dict = idealized_aa.residue_dict()
self.n_run = n_run
if self.r is None:
self.r = rama_eval()
self.rotamer_manager = rotamer_manager
if self.rotamer_manager is None:
self.rotamer_manager = RotamerEval()
ram = ramalyze.ramalyze(pdb_hierarchy=pdb_hierarchy)
self.p_initial_rama_outliers = ram.out_percent
self.p_before_minimization_rama_outliers = None
self.p_after_minimiaztion_rama_outliers = None
n_inputs = [reference_map, crystal_symmetry].count(None)
if not (n_inputs == 0 or n_inputs == 2):
print >> log, "Need to have both map and symmetry info. Not using map."
self.reference_map = None
# here we are recording what CCD solutions were used to fix particular
# outliers to not use the same in the next CCD try.
# Nested dict. First level:
# key: chain id, value: dict
# key: resid (string), value: list of tried variants.
self.tried_rama_angles = tried_rama_angles
self.tried_final_rama_angles = tried_final_rama_angles
berkeley_count = utils.list_rama_outliers_h(self.resulting_pdb_h).count("\n")
self.berkeley_p_before_minimization_rama_outliers = \
berkeley_count/float(self.resulting_pdb_h.overall_counts().n_residues)*100
n_bad_omegas = utils.n_bad_omegas(self.resulting_pdb_h)
self.berkeley_p_after_minimiaztion_rama_outliers = self.berkeley_p_before_minimization_rama_outliers
self.ref_exclusion_selection = ""
self.number_of_ccd_trials = 0
# print "logic expr outcome:", (self.number_of_ccd_trials < 10 and self.berkeley_p_before_minimization_rama_outliers > 0.001)
# print self.number_of_ccd_trials < 10
# print "berkeley before rama out:", self.berkeley_p_before_minimization_rama_outliers
if (self.berkeley_p_before_minimization_rama_outliers <= 0.001 and
(n_bad_omegas<1 and self.params.make_all_trans)):
print >> self.log, "No ramachandran outliers, skipping CCD step."
print "n_bad_omegas", n_bad_omegas
print "self.params.make_all_trans",self.params.make_all_trans
if not self.params.enabled:
print >> self.log, "Loop idealization is not enabled, use 'enabled=True'."
while (self.number_of_ccd_trials < self.params.number_of_ccd_trials
and (self.berkeley_p_after_minimiaztion_rama_outliers > 0.001 or
(n_bad_omegas>=1 and self.params.make_all_trans))
and self.params.enabled):
print >> self.log, "CCD try number, outliers:", self.number_of_ccd_trials, self.berkeley_p_before_minimization_rama_outliers
processed_chain_ids = []
for chain in self.resulting_pdb_h.only_model().chains():
if chain.id not in self.tried_rama_angles.keys():
self.tried_rama_angles[chain.id] = {}
if chain.id not in self.tried_final_rama_angles.keys():
self.tried_final_rama_angles[chain.id] = {}
print >> self.log, "Idealizing chain %s" % chain.id
if chain.id not in processed_chain_ids:
processed_chain_ids.append(chain.id)
else:
continue
selection = "protein and chain %s and (name N or name CA or name C or name O)" % chain.id
sel = asc.selection("chain %s" % chain.id)
chain_h = self.resulting_pdb_h.select(sel)
m = chain_h.only_model()
i = 0
cutted_chain_h = None
for c in m.chains():
if i == 0:
cutted_chain_h = iotbx.pdb.hierarchy.new_hierarchy_from_chain(c)
else:
print >> self.log, "WARNING!!! Duplicating chain ids! Only the first chain will be processed."
print >> self.log, " Removing chain %s with %d residues" % (c.id, len(c.residues()))
m.remove_chain(c)
i += 1
exclusions, ch_h = self.idealize_chain(
hierarchy=(cutted_chain_h if cutted_chain_h else chain_h),
tried_rama_angles_for_chain=self.tried_rama_angles[chain.id],
tried_final_rama_angles_for_chain=self.tried_final_rama_angles[chain.id])
if ch_h is not None:
set_xyz_smart(
# dest_h=self.resulting_pdb_h,
dest_h=chain,
source_h=ch_h)
for resnum in exclusions:
selection += " and not resseq %s" % resnum
self.ref_exclusion_selection += "(%s) or " % selection
print "self.tried_rama_angles", self.tried_rama_angles
print "self.tried_final_rama_angles", self.tried_final_rama_angles
#
# dumping and reloading hierarchy to do proper rounding of coordinates
self.resulting_pdb_h = iotbx.pdb.input(
source_info=None,
lines=self.resulting_pdb_h.as_pdb_string()).construct_hierarchy()
berkeley_count = utils.list_rama_outliers_h(self.resulting_pdb_h).count("\n")
self.berkeley_p_before_minimization_rama_outliers = \
berkeley_count/float(self.resulting_pdb_h.overall_counts().n_residues)*100
if len(self.ref_exclusion_selection) > 0:
self.ref_exclusion_selection = self.ref_exclusion_selection[:-3]
ram = ramalyze.ramalyze(pdb_hierarchy=self.resulting_pdb_h)
self.p_before_minimization_rama_outliers = ram.out_percent
duke_count = ram.get_outliers_count_and_fraction()[0]
if berkeley_count != duke_count:
print >> self.log, "Discrepancy between berkeley and duke after ccd:", berkeley_count, duke_count
self.resulting_pdb_h.write_pdb_file(file_name="%d%s_discrepancy.pdb" % (self.number_of_ccd_trials, self.params.output_prefix))
if self.params.debug:
self.resulting_pdb_h.write_pdb_file(
file_name="%d%s_all_not_minized.pdb" % (self.number_of_ccd_trials,
self.params.output_prefix))
if self.params.minimize_whole:
print >> self.log, "minimizing whole chain..."
print >> self.log, "self.ref_exclusion_selection", self.ref_exclusion_selection
# print >> sel
# XXX but first let's check and fix rotamers...
print >> self.log, "Fixing/checking rotamers in loop idealization..."
excl_sel = self.ref_exclusion_selection
if len(excl_sel) == 0:
excl_sel = None
non_outliers_for_check = asc.selection("(%s)" % self.ref_exclusion_selection)
pre_result_h = mmtbx.utils.fix_rotamer_outliers(
pdb_hierarchy=self.resulting_pdb_h,
grm=self.grm.geometry,
xrs=self.xrs,
map_data=self.reference_map,
radius=5,
mon_lib_srv=None,
rotamer_manager=self.rotamer_manager,
backrub_range=None, # don't sample backrub at this point
non_outliers_to_check=non_outliers_for_check, # bool selection
asc=asc,
verbose=True,
log=self.log)
if self.reference_map is None:
minimize_wrapper_for_ramachandran(
hierarchy=self.resulting_pdb_h,
xrs=self.xrs,
original_pdb_h=self.original_pdb_h,
excl_string_selection=self.ref_exclusion_selection,
grm=self.grm,
log=None,
ss_annotation=self.secondary_structure_annotation)
else:
mwwm = minimize_wrapper_with_map(
pdb_h=self.resulting_pdb_h,
xrs=self.xrs,
target_map=self.reference_map,
grm=self.grm,
ss_annotation=self.secondary_structure_annotation,
number_of_cycles=Auto,
log=self.log)
if self.params.debug:
self.resulting_pdb_h.write_pdb_file(
file_name="%d%s_all_minized.pdb" % (self.number_of_ccd_trials,
self.params.output_prefix))
ram = ramalyze.ramalyze(pdb_hierarchy=self.resulting_pdb_h)
self.p_after_minimiaztion_rama_outliers = ram.out_percent
berkeley_count = utils.list_rama_outliers_h(self.resulting_pdb_h).count("\n")
duke_count = ram.get_outliers_count_and_fraction()[0]
n_bad_omegas = utils.n_bad_omegas(self.resulting_pdb_h)
self.berkeley_p_after_minimiaztion_rama_outliers = \
berkeley_count/float(self.resulting_pdb_h.overall_counts().n_residues)*100
if berkeley_count != duke_count:
print >> self.log, "Discrepancy between berkeley and duke after min:", berkeley_count, duke_count
else:
print >> self.log, "Number of Rama outliers after min:", berkeley_count
print >> self.log, "Number of bad omegas:", n_bad_omegas
self.number_of_ccd_trials += 1
def backrub_move(
prev_res,
cur_res,
next_res,
angle,
move_oxygens=False,
accept_worse_rama=False,
rotamer_manager=None,
rama_manager=None):
import boost.python
ext = boost.python.import_ext("mmtbx_validation_ramachandran_ext")
from mmtbx_validation_ramachandran_ext import rama_eval
from scitbx.matrix import rotate_point_around_axis
from mmtbx.conformation_dependent_library.multi_residue_class import ThreeProteinResidues, \
RestraintsRegistry
if abs(angle) < 1e-4:
return
if prev_res is None or next_res is None:
return
saved_res = [{},{},{}]
for i, r in enumerate([prev_res, cur_res, next_res]):
for a in r.atoms():
saved_res[i][a.name.strip()] = a.xyz
if rotamer_manager is None:
rotamer_manager = RotamerEval()
prev_ca = prev_res.find_atom_by(name=" CA ")
cur_ca = cur_res.find_atom_by(name=" CA ")
next_ca = next_res.find_atom_by(name=" CA ")
if prev_ca is None or next_ca is None or cur_ca is None:
return
atoms_to_move = []
atoms_to_move.append(prev_res.find_atom_by(name=" C "))
atoms_to_move.append(prev_res.find_atom_by(name=" O "))
for atom in cur_res.atoms():
atoms_to_move.append(atom)
atoms_to_move.append(next_res.find_atom_by(name=" N "))
for atom in atoms_to_move:
assert atom is not None
new_xyz = rotate_point_around_axis(
axis_point_1 = prev_ca.xyz,
axis_point_2 = next_ca.xyz,
point = atom.xyz,
angle = angle,
deg = True)
atom.xyz = new_xyz
if move_oxygens:
registry = RestraintsRegistry()
if rama_manager is None:
rama_manager = rama_eval()
tpr = ThreeProteinResidues(geometry=None, registry=registry)
tpr.append(prev_res)
tpr.append(cur_res)
tpr.append(next_res)
phi_psi_angles = tpr.get_phi_psi_angles()
rama_key = tpr.get_ramalyze_key()
ev_before = rama_manager.evaluate_angles(rama_key, phi_psi_angles[0], phi_psi_angles[1])
theta1 = _find_theta(
ap1 = prev_ca.xyz,
ap2 = cur_ca.xyz,
cur_xyz = prev_res.find_atom_by(name=" O ").xyz,
needed_xyz = saved_res[0]["O"])
theta2 = _find_theta(
ap1 = cur_ca.xyz,
ap2 = next_ca.xyz,
cur_xyz = cur_res.find_atom_by(name=" O ").xyz,
needed_xyz = saved_res[1]["O"])
for a in [prev_res.find_atom_by(name=" C "),
prev_res.find_atom_by(name=" O "),
cur_res.find_atom_by(name=" C ")]:
new_xyz = rotate_point_around_axis(
axis_point_1 = prev_ca.xyz,
axis_point_2 = cur_ca.xyz,
point = a.xyz,
angle = theta1,
deg = True)
a.xyz = new_xyz
for a in [cur_res.find_atom_by(name=" C "),
cur_res.find_atom_by(name=" O "),
next_res.find_atom_by(name=" N ")]:
new_xyz = rotate_point_around_axis(
axis_point_1 = cur_ca.xyz,
axis_point_2 = next_ca.xyz,
point = a.xyz,
angle = theta2,
deg = True)
a.xyz = new_xyz
phi_psi_angles = tpr.get_phi_psi_angles()
rama_key = tpr.get_ramalyze_key()
ev_after = rama_manager.evaluate_angles(rama_key, phi_psi_angles[0], phi_psi_angles[1])
if ev_before > ev_after and not accept_worse_rama:
for a in [prev_res.find_atom_by(name=" C "),
prev_res.find_atom_by(name=" O "),
cur_res.find_atom_by(name=" C ")]:
new_xyz = rotate_point_around_axis(
axis_point_1 = prev_ca.xyz,
axis_point_2 = cur_ca.xyz,
point = a.xyz,
angle = -theta1,
deg = True)
a.xyz = new_xyz
for a in [cur_res.find_atom_by(name=" C "),
cur_res.find_atom_by(name=" O "),
next_res.find_atom_by(name=" N ")]:
new_xyz = rotate_point_around_axis(
axis_point_1 = cur_ca.xyz,
axis_point_2 = next_ca.xyz,
point = a.xyz,
angle = -theta2,
deg = True)
a.xyz = new_xyz
def __init__(self,
pdb_hierarchy,
params=None,
secondary_structure_annotation=None,
log=null_out(),
verbose=True):
if len(pdb_hierarchy.models()) > 1:
raise Sorry("Multi-model files are not supported")
self.original_pdb_h = pdb_hierarchy
self.secondary_structure_annotation=secondary_structure_annotation
xrs = pdb_hierarchy.extract_xray_structure()
asc = pdb_hierarchy.atom_selection_cache()
self.resulting_pdb_h = pdb_hierarchy.deep_copy()
self.resulting_pdb_h.reset_atom_i_seqs()
self.params = self.process_params(params)
self.log = log
self.verbose = verbose
self.r = rama_eval()
self.rotamer_manager = RotamerEval()
ram = ramalyze.ramalyze(pdb_hierarchy=pdb_hierarchy)
self.p_initial_rama_outliers = ram.out_percent
self.p_before_minimization_rama_outliers = None
self.p_after_minimiaztion_rama_outliers = None
berkeley_count = utils.list_rama_outliers_h(self.resulting_pdb_h).count("\n")
self.berkeley_p_before_minimization_rama_outliers = \
berkeley_count/float(self.resulting_pdb_h.overall_counts().n_residues)*100
# self.berkeley_p_before_minimization_rama_outliers = None
self.berkeley_p_after_minimiaztion_rama_outliers = None
self.ref_exclusion_selection = ""
number_of_ccd_trials = 0
# print "logic expr outcome:", (number_of_ccd_trials < 10 and self.berkeley_p_before_minimization_rama_outliers > 0.001)
# print number_of_ccd_trials < 10
# print "berkeley before rama out:", self.berkeley_p_before_minimization_rama_outliers
if self.berkeley_p_before_minimization_rama_outliers <= 0.001:
print >> self.log, "No ramachandran outliers, skipping CCD step."
if not self.params.enabled:
print >> self.log, "Loop idealization is not enabled, use 'enabled=True'."
while (number_of_ccd_trials < self.params.number_of_ccd_trials
and self.berkeley_p_before_minimization_rama_outliers > 0.001
and self.params.enabled):
print "CCD try number, outliers:", number_of_ccd_trials, self.berkeley_p_before_minimization_rama_outliers
number_of_ccd_trials += 1
processed_chain_ids = []
for chain in self.resulting_pdb_h.only_model().chains():
print >> self.log, "Idealizing chain %s" % chain.id
if chain.id not in processed_chain_ids:
processed_chain_ids.append(chain.id)
else:
continue
selection = "protein and chain %s and (name N or name CA or name C or name O)" % chain.id
sel = asc.selection("chain %s" % chain.id)
chain_h = self.resulting_pdb_h.select(sel)
m = chain_h.only_model()
i = 0
cutted_chain_h = None
for c in m.chains():
if i == 0:
cutted_chain_h = iotbx.pdb.hierarchy.new_hierarchy_from_chain(c)
else:
print >> self.log, "WARNING!!! Duplicating chain ids! Only the first chain will be processed."
print >> self.log, " Removing chain %s with %d residues" % (c.id, len(c.residues()))
m.remove_chain(c)
i += 1
exclusions, ch_h = self.idealize_chain(
hierarchy=(cutted_chain_h if cutted_chain_h else chain_h))
if ch_h is not None:
set_xyz_smart(
# dest_h=self.resulting_pdb_h,
dest_h=chain,
source_h=ch_h)
for resnum in exclusions:
selection += " and not resseq %s" % resnum
self.ref_exclusion_selection += "(%s) or " % selection
#
# dumping and reloading hierarchy to do proper rounding of coordinates
self.resulting_pdb_h = iotbx.pdb.input(
source_info=None,
lines=self.resulting_pdb_h.as_pdb_string()).construct_hierarchy()
berkeley_count = utils.list_rama_outliers_h(self.resulting_pdb_h).count("\n")
self.berkeley_p_before_minimization_rama_outliers = \
berkeley_count/float(self.resulting_pdb_h.overall_counts().n_residues)*100
if len(self.ref_exclusion_selection) > 0:
self.ref_exclusion_selection = self.ref_exclusion_selection[:-3]
self.resulting_pdb_h.write_pdb_file(file_name="%s_before_minimization.pdb" % self.params.output_prefix)
ram = ramalyze.ramalyze(pdb_hierarchy=self.resulting_pdb_h)
self.p_before_minimization_rama_outliers = ram.out_percent
duke_count = ram.get_outliers_count_and_fraction()[0]
if berkeley_count != duke_count:
print >> self.log, "Discrepancy between berkeley and duke after ccd:", berkeley_count, duke_count
if self.params.minimize_whole:
print >> self.log, "minimizing whole thing..."
print >> self.log, "self.ref_exclusion_selection", self.ref_exclusion_selection
# print >> sel
minimize_wrapper_for_ramachandran(
hierarchy=self.resulting_pdb_h,
xrs=xrs,
original_pdb_h=self.original_pdb_h,
excl_string_selection=self.ref_exclusion_selection,
log=None,
ss_annotation=self.secondary_structure_annotation)
# self.resulting_pdb_h.write_pdb_file(file_name="%s_all_minized.pdb" % self.params.output_prefix)
ram = ramalyze.ramalyze(pdb_hierarchy=self.resulting_pdb_h)
self.p_after_minimiaztion_rama_outliers = ram.out_percent
berkeley_count = utils.list_rama_outliers_h(self.resulting_pdb_h).count("\n")
duke_count = ram.get_outliers_count_and_fraction()[0]
self.berkeley_p_after_minimiaztion_rama_outliers = \
berkeley_count/float(self.resulting_pdb_h.overall_counts().n_residues)*100
if berkeley_count != duke_count:
print >> self.log, "Discrepancy between berkeley and duke after min:", berkeley_count, duke_count
else:
print >> self.log, "Number of Rama outliers after min:", berkeley_count
def extract_proxies(self, hierarchy):
favored = ramalyze.RAMALYZE_FAVORED
allowed = ramalyze.RAMALYZE_ALLOWED
outlier = ramalyze.RAMALYZE_OUTLIER
self.hierarchy = hierarchy
bool_atom_selection = self._determine_bool_atom_selection(hierarchy)
selected_h = hierarchy.select(bool_atom_selection)
n_seq = flex.max(selected_h.atoms().extract_i_seq())
# Drop all previous proxies
self._oldfield_proxies = ext.shared_phi_psi_proxy()
self._emsley_proxies = ext.shared_phi_psi_proxy()
self._emsley8k_proxies = ext.shared_phi_psi_proxy()
# it would be great to save rama_eval, but the fact that this is called in
# pdb_interpretation, not in mmtbx.model makes it impossible
self.rama_eval = rama_eval()
for three in generate_protein_threes(hierarchy=selected_h,
geometry=None):
rc = three.get_phi_psi_atoms()
if rc is None: continue
rama_key = three.get_ramalyze_key()
angles = three.get_phi_psi_angles()
rama_score = self.rama_eval.get_score(rama_key, angles[0],
angles[1])
r_eval = self.rama_eval.evaluate_score(rama_key, rama_score)
phi_atoms, psi_atoms = rc
i_seqs = [atom.i_seq for atom in phi_atoms] + [psi_atoms[-1].i_seq]
resnames = three.get_resnames()
r_name = resnames[1]
assert rama_key in range(6)
text_rama_key = ramalyze.res_types[rama_key]
assert text_rama_key in [
"general", "glycine", "cis-proline", "trans-proline",
"pre-proline", "isoleucine or valine"
]
# pick where to put...
ev_match_dict = {
favored: self.params.favored,
allowed: self.params.allowed,
outlier: self.params.outlier
}
r_type = ev_match_dict[r_eval]
if r_type == 'oldfield':
proxy = ext.phi_psi_proxy(residue_type=text_rama_key,
i_seqs=i_seqs,
weight=1) # XXX Not used in oldfield
self.append_oldfield_proxies(proxy, n_seq)
### THIS IS CRUEL. REMOVE ONCE favored/allowed/outlier are made multiple!
if (self.params.inject_emsley8k_into_oldfield_favored):
proxy = ext.phi_psi_proxy(residue_type=text_rama_key,
i_seqs=i_seqs,
weight=5)
self.append_emsley8k_proxies(proxy, n_seq)
###
elif r_type == 'emsley':
weight = self.params.emsley.weight
proxy = ext.phi_psi_proxy(residue_type=text_rama_key,
i_seqs=i_seqs,
weight=weight)
self.append_emsley_proxies(proxy, n_seq)
elif r_type == 'emsley8k':
if (r_eval is favored):
weight = self.params.emsley8k.weight_favored
elif (r_eval is allowed):
weight = self.params.emsley8k.weight_allowed
elif (r_eval is outlier):
weight = self.params.emsley8k.weight_outlier
else:
raise RuntimeError("Rama eveluation failed.")
proxy = ext.phi_psi_proxy(residue_type=text_rama_key,
i_seqs=i_seqs,
weight=weight)
self.append_emsley8k_proxies(proxy, n_seq)
elif (r_type is None):
pass
else:
raise RuntimeError("Not an option: %s" % str(r_type))
print("", file=self.log)
print(" %d Ramachandran restraints generated." %
(self.get_n_proxies()),
file=self.log)
print(" %d Oldfield and %d Emsley and %d emsley8k." %
(self.get_n_oldfield_proxies(), self.get_n_emsley_proxies(),
self.get_n_emsley8k_proxies()),
file=self.log)
def __init__(self,
pdb_hierarchy,
params=None,
secondary_structure_annotation=None,
reference_map=None,
crystal_symmetry=None,
grm=None,
rama_manager=None,
rotamer_manager=None,
log=null_out(),
verbose=False):
if len(pdb_hierarchy.models()) > 1:
raise Sorry("Multi-model files are not supported")
self.original_pdb_h = pdb_hierarchy
self.secondary_structure_annotation = secondary_structure_annotation
asc = pdb_hierarchy.atom_selection_cache()
self.xrs = pdb_hierarchy.extract_xray_structure(
crystal_symmetry=crystal_symmetry)
self.reference_map = reference_map
self.resulting_pdb_h = pdb_hierarchy.deep_copy()
self.resulting_pdb_h.reset_atom_i_seqs()
self.params = self.process_params(params)
self.log = log
self.verbose = verbose
self.grm = grm
self.r = rama_manager
if self.r is None:
self.r = rama_eval()
self.rotamer_manager = rotamer_manager
if self.rotamer_manager is None:
self.rotamer_manager = RotamerEval()
ram = ramalyze.ramalyze(pdb_hierarchy=pdb_hierarchy)
self.p_initial_rama_outliers = ram.out_percent
self.p_before_minimization_rama_outliers = None
self.p_after_minimiaztion_rama_outliers = None
n_inputs = [reference_map, crystal_symmetry].count(None)
if not (n_inputs == 0 or n_inputs == 2):
print >> log, "Need to have both map and symmetry info. Not using map."
self.reference_map = None
berkeley_count = utils.list_rama_outliers_h(
self.resulting_pdb_h).count("\n")
self.berkeley_p_before_minimization_rama_outliers = \
berkeley_count/float(self.resulting_pdb_h.overall_counts().n_residues)*100
# self.berkeley_p_before_minimization_rama_outliers = None
self.berkeley_p_after_minimiaztion_rama_outliers = self.berkeley_p_before_minimization_rama_outliers
self.ref_exclusion_selection = ""
number_of_ccd_trials = 0
# print "logic expr outcome:", (number_of_ccd_trials < 10 and self.berkeley_p_before_minimization_rama_outliers > 0.001)
# print number_of_ccd_trials < 10
# print "berkeley before rama out:", self.berkeley_p_before_minimization_rama_outliers
if self.berkeley_p_before_minimization_rama_outliers <= 0.001:
print >> self.log, "No ramachandran outliers, skipping CCD step."
if not self.params.enabled:
print >> self.log, "Loop idealization is not enabled, use 'enabled=True'."
while (number_of_ccd_trials < self.params.number_of_ccd_trials
and self.berkeley_p_after_minimiaztion_rama_outliers > 0.001
and self.params.enabled):
print "CCD try number, outliers:", number_of_ccd_trials, self.berkeley_p_before_minimization_rama_outliers
number_of_ccd_trials += 1
processed_chain_ids = []
for chain in self.resulting_pdb_h.only_model().chains():
print >> self.log, "Idealizing chain %s" % chain.id
if chain.id not in processed_chain_ids:
processed_chain_ids.append(chain.id)
else:
continue
selection = "protein and chain %s and (name N or name CA or name C or name O)" % chain.id
sel = asc.selection("chain %s" % chain.id)
chain_h = self.resulting_pdb_h.select(sel)
m = chain_h.only_model()
i = 0
cutted_chain_h = None
for c in m.chains():
if i == 0:
cutted_chain_h = iotbx.pdb.hierarchy.new_hierarchy_from_chain(
c)
else:
print >> self.log, "WARNING!!! Duplicating chain ids! Only the first chain will be processed."
print >> self.log, " Removing chain %s with %d residues" % (
c.id, len(c.residues()))
m.remove_chain(c)
i += 1
exclusions, ch_h = self.idealize_chain(
hierarchy=(cutted_chain_h if cutted_chain_h else chain_h))
if ch_h is not None:
set_xyz_smart(
# dest_h=self.resulting_pdb_h,
dest_h=chain,
source_h=ch_h)
for resnum in exclusions:
selection += " and not resseq %s" % resnum
self.ref_exclusion_selection += "(%s) or " % selection
#
# dumping and reloading hierarchy to do proper rounding of coordinates
self.resulting_pdb_h = iotbx.pdb.input(
source_info=None, lines=self.resulting_pdb_h.as_pdb_string(
)).construct_hierarchy()
berkeley_count = utils.list_rama_outliers_h(
self.resulting_pdb_h).count("\n")
self.berkeley_p_before_minimization_rama_outliers = \
berkeley_count/float(self.resulting_pdb_h.overall_counts().n_residues)*100
if len(self.ref_exclusion_selection) > 0:
self.ref_exclusion_selection = self.ref_exclusion_selection[:
-3]
# self.resulting_pdb_h.write_pdb_file(file_name="%s_before_minimization.pdb" % self.params.output_prefix)
ram = ramalyze.ramalyze(pdb_hierarchy=self.resulting_pdb_h)
self.p_before_minimization_rama_outliers = ram.out_percent
duke_count = ram.get_outliers_count_and_fraction()[0]
if berkeley_count != duke_count:
print >> self.log, "Discrepancy between berkeley and duke after ccd:", berkeley_count, duke_count
if self.params.minimize_whole:
print >> self.log, "minimizing whole thing..."
print >> self.log, "self.ref_exclusion_selection", self.ref_exclusion_selection
# print >> sel
if self.reference_map is None:
minimize_wrapper_for_ramachandran(
hierarchy=self.resulting_pdb_h,
xrs=self.xrs,
original_pdb_h=self.original_pdb_h,
excl_string_selection=self.ref_exclusion_selection,
grm=self.grm,
log=None,
ss_annotation=self.secondary_structure_annotation)
else:
mwwm = minimize_wrapper_with_map(
pdb_h=self.resulting_pdb_h,
xrs=self.xrs,
target_map=self.reference_map,
grm=self.grm,
ss_annotation=self.secondary_structure_annotation,
log=self.log)
# self.resulting_pdb_h.write_pdb_file(file_name="%s_all_minized.pdb" % self.params.output_prefix)
ram = ramalyze.ramalyze(pdb_hierarchy=self.resulting_pdb_h)
self.p_after_minimiaztion_rama_outliers = ram.out_percent
berkeley_count = utils.list_rama_outliers_h(
self.resulting_pdb_h).count("\n")
duke_count = ram.get_outliers_count_and_fraction()[0]
self.berkeley_p_after_minimiaztion_rama_outliers = \
berkeley_count/float(self.resulting_pdb_h.overall_counts().n_residues)*100
if berkeley_count != duke_count:
print >> self.log, "Discrepancy between berkeley and duke after min:", berkeley_count, duke_count
else:
print >> self.log, "Number of Rama outliers after min:", berkeley_count
def get_all_starting_conformations(moving_h,
change_radius,
include_allowed,
n_outliers,
direction_forward=True,
cutoff=50,
change_all=True,
log=null_out(),
check_omega=False):
if log is None:
log = StringIO()
variants = []
result = []
r = rama_eval()
phi_psi_atoms = utils.get_phi_psi_atoms(moving_h, omega=True)
# print "N residue groups in h", [x.resseq for x in moving_h.residue_groups()]
if len(phi_psi_atoms) == 0:
print("Strange input to starting conformations!!!", file=log)
return result
n_rama = len(phi_psi_atoms)
# print "n_rama", n_rama
change_angles = [None]
if change_all:
change_angles = range(
(n_rama) // 2 - change_radius - n_outliers // 2,
(n_rama) // 2 + change_radius + 1 + n_outliers // 2)
# if change_angles[0] < 0:
# change_angles = range(change_angles[-1]-change_angles[0])
has_twisted = False
if check_omega:
omegas = [x[2] for x in phi_psi_atoms]
for o in omegas:
if o is not None and abs(abs(o) - 180) > 30:
has_twisted = True
print("n_outliers", n_outliers, file=log)
for i, (phi_psi_pair, rama_key, omega) in enumerate(phi_psi_atoms):
angle_is_outlier = utils.rama_evaluate(
phi_psi_pair, r, rama_key) == ramalyze.RAMALYZE_OUTLIER
angle_is_outlier = angle_is_outlier or (
include_allowed and utils.rama_evaluate(
phi_psi_pair, r, rama_key) == ramalyze.RAMALYZE_ALLOWED)
twisted = omega is not None and ((abs(abs(omega) - 180) > 30)
and check_omega)
print("in cycle, N, outlier?, change?, twisted?",
i,
angle_is_outlier,
i in change_angles,
twisted,
file=log)
if angle_is_outlier and n_outliers < 3:
vs = get_sampled_rama_favored_angles(rama_key, r)
elif (i in change_angles) or angle_is_outlier or has_twisted:
# vs = get_sampled_rama_favored_angles(rama_key, r)
vs = ramalyze.get_favored_regions(rama_key)
else:
vs = [(None, None)]
variants.append(vs)
print("variants", variants, file=log)
# Filtering them, since could be
# [len(x) for x in variants] = [129, 129, 4, 129, 129]
# resulting in 1107691524 all_angles_combination
n_comb = numpy.prod([len(x) for x in variants])
if n_comb > cutoff:
# still aiming for ~1000
n_in_each = int(1000**(1 / len(variants)))
variants = [
random.sample(x, n_in_each) if len(x) > n_in_each else x
for x in variants
]
all_angles_combination = list(itertools.product(*variants))
# filter none combinations
# print "len(all_angles_combination)", len(all_angles_combination)
all_angles_combination_f = []
for comb in all_angles_combination:
if is_not_none_combination(comb):
all_angles_combination_f.append(comb)
print("len(all_angles_combination_f)",
len(all_angles_combination_f),
file=log)
return all_angles_combination_f