def __init__(self,
data_arrays,
xray_structure,
log=None,
silent=False,
output_file=None,
peak_search=False,
map_cutoff=None,
peak_search_params=None,
r_free_arrays=None,
write_map=True,
multiscale=False,
anomalous=False):
if (log is None): log = sys.stdout
adopt_init_args(self, locals())
fmodels = []
for i_seq, d in enumerate(data_arrays):
if (not silent):
print("Data set: %d" % i_seq, file=log)
if (d.anomalous_flag()) and (not anomalous):
d = d.average_bijvoet_mates()
elif (anomalous):
assert d.anomalous_flag()
if (r_free_arrays is not None) and (i_seq < len(r_free_arrays)):
r_free_flags = r_free_arrays[i_seq]
else:
r_free_flags = d.array(data=flex.bool(d.data().size(), False))
fmodel = mmtbx.f_model.manager(xray_structure=xray_structure,
r_free_flags=r_free_flags,
target_name="ls_wunit_k1",
f_obs=d)
fmodel.update_all_scales(log=None)
if (not silent):
fmodel.info().show_rfactors_targets_scales_overall(out=log)
print(file=log)
fmodels.append(fmodel)
self.fmodel = fmodels[0]
# prepare Fobs for map calculation (apply scaling):
f_obss = []
for fmodel in fmodels:
obs = fmodel.f_obs()
f_obs_scale = 1.0 / fmodel.k_anisotropic() / fmodel.k_isotropic()
obs = miller.array(miller_set=fmodel.f_model(),
data=obs.data() * f_obs_scale)
f_obss.append(obs)
# given two Fobs sets, make them one-to-one matching, get phases and map coefficients
# Note: f_calc below is just f_calc from atoms (no bulk solvent etc applied)
fobs_1, f_model = f_obss[0].common_sets(other=fmodels[1].f_model())
fobs_1, fobs_2 = fobs_1.common_sets(other=f_obss[1])
fobs_1, f_model = fobs_1.common_sets(other=f_model)
self.f_model = f_model
assert fobs_2.indices().all_eq(fobs_1.indices())
assert f_model.indices().all_eq(fobs_1.indices())
# scale again
scale_k1 = 1
den = flex.sum(flex.abs(fobs_2.data()) * flex.abs(fobs_2.data()))
if (den != 0):
scale_k1 = flex.sum(
flex.abs(fobs_1.data()) * flex.abs(fobs_2.data())) / den
#
fobs_2 = fobs_2.array(data=fobs_2.data() * scale_k1)
if multiscale:
fobs_1 = fobs_2.multiscale(other=fobs_1, reflections_per_bin=250)
if (not silent):
print("", file=log)
print("Fobs1_vs_Fobs2 statistics:", file=log)
print("Bin# Resolution range Compl. No.of refl. CC R-factor",
file=log)
fobs_1.setup_binner(
reflections_per_bin=min(500,
fobs_1.data().size()))
fobs_2.use_binning_of(fobs_1)
for i_bin in fobs_1.binner().range_used():
sel = fobs_1.binner().selection(i_bin)
f1 = fobs_1.select(sel)
f2 = fobs_2.select(sel)
d_max, d_min = fobs_1.d_max_min()
compl = fobs_1.completeness(d_max=d_max)
n_ref = sel.count(True)
num = flex.sum(flex.abs(f1.data() - f2.data()))
den = flex.sum(flex.abs(f1.data() + f2.data()) / 2)
r = None
if (den != 0):
r = num / den
cc = flex.linear_correlation(x=f1.data(),
y=f2.data()).coefficient()
d_range = fobs_1.binner().bin_legend(i_bin=i_bin,
show_bin_number=False,
show_counts=False)
fmt = "%3d: %-17s %4.2f %6d %5.3f %6s"
print(fmt % (i_bin, d_range, compl, n_ref, cc,
format_value("%6.4f", r)),
file=log)
# overall statistics
self.cc = flex.linear_correlation(x=fobs_1.data(),
y=fobs_2.data()).coefficient()
num = flex.sum(flex.abs(fobs_1.data() - fobs_2.data()))
den = flex.sum(flex.abs(fobs_2.data() + fobs_2.data()) / 2)
self.r_factor = None
if (den != 0):
self.r_factor = num / den
# map coefficients
def phase_transfer(miller_array, phase_source):
tmp = miller.array(
miller_set=miller_array,
data=flex.double(miller_array.indices().size(),
1)).phase_transfer(phase_source=phase_source)
return miller.array(miller_set=miller_array,
data=miller_array.data() * tmp.data())
if (not anomalous):
diff = miller.array(miller_set=f_model,
data=fobs_1.data() - fobs_2.data())
self.map_coeff = phase_transfer(miller_array=diff,
phase_source=f_model)
else:
dano_1 = fobs_1.anomalous_differences()
dano_2 = fobs_2.anomalous_differences()
assert dano_1.indices().all_eq(dano_2.indices())
diff = miller.array(miller_set=dano_1,
data=dano_1.data() - dano_2.data())
f_model_phases = f_model.average_bijvoet_mates().common_set(diff)
map_coeffs = phase_transfer(miller_array=diff,
phase_source=f_model_phases)
self.map_coeff = map_coeffs.customized_copy(
data=map_coeffs.data() / (2j))
if (self.map_coeff.anomalous_flag()):
self.map_coeff = map_coeff.average_bijvoet_mates()
self.file_names = []
if (write_map):
self.file_names = self.write_map_file()
def __init__ (self,
data_arrays,
xray_structure,
log=None,
silent=False,
output_file=None,
peak_search=False,
map_cutoff=None,
peak_search_params=None,
r_free_arrays=None,
write_map=True,
multiscale=False,
anomalous=False) :
if (log is None) : log = sys.stdout
adopt_init_args(self, locals())
fmodels = []
for i_seq, d in enumerate(data_arrays):
if(not silent):
print >> log, "Data set: %d"%i_seq
if(d.anomalous_flag()) and (not anomalous) :
d = d.average_bijvoet_mates()
elif (anomalous) :
assert d.anomalous_flag()
if (r_free_arrays is not None) and (i_seq < len(r_free_arrays)) :
r_free_flags = r_free_arrays[i_seq]
else :
r_free_flags = d.array(data = flex.bool(d.data().size(), False))
fmodel = mmtbx.f_model.manager(
xray_structure = xray_structure,
r_free_flags = r_free_flags,
target_name = "ls_wunit_k1",
f_obs = d)
fmodel.update_all_scales(log=None)
if(not silent):
fmodel.info().show_rfactors_targets_scales_overall(out=log)
print >> log
fmodels.append(fmodel)
self.fmodel = fmodels[0]
# prepare Fobs for map calculation (apply scaling):
f_obss = []
for fmodel in fmodels:
obs = fmodel.f_obs()
f_obs_scale = 1.0 / fmodel.k_anisotropic() / fmodel.k_isotropic()
obs = miller.array(miller_set = fmodel.f_model(),
data = obs.data()*f_obs_scale)
f_obss.append(obs)
# given two Fobs sets, make them one-to-one matching, get phases and map coefficients
# Note: f_calc below is just f_calc from atoms (no bulk solvent etc applied)
fobs_1, f_model = f_obss[0].common_sets(other = fmodels[1].f_model())
fobs_1, fobs_2 = fobs_1.common_sets(other = f_obss[1])
fobs_1, f_model = fobs_1.common_sets(other = f_model)
self.f_model = f_model
assert fobs_2.indices().all_eq(fobs_1.indices())
assert f_model.indices().all_eq(fobs_1.indices())
# scale again
scale_k1 = 1
den = flex.sum(flex.abs(fobs_2.data())*flex.abs(fobs_2.data()))
if(den != 0):
scale_k1 = flex.sum(flex.abs(fobs_1.data())*flex.abs(fobs_2.data())) / den
#
fobs_2 = fobs_2.array(data = fobs_2.data()*scale_k1)
if multiscale:
fobs_1 = fobs_2.multiscale(other = fobs_1, reflections_per_bin=250)
if(not silent):
print >> log, ""
print >> log, "Fobs1_vs_Fobs2 statistics:"
print >> log, "Bin# Resolution range Compl. No.of refl. CC R-factor"
fobs_1.setup_binner(reflections_per_bin = min(500, fobs_1.data().size()))
fobs_2.use_binning_of(fobs_1)
for i_bin in fobs_1.binner().range_used():
sel = fobs_1.binner().selection(i_bin)
f1 = fobs_1.select(sel)
f2 = fobs_2.select(sel)
d_max, d_min = fobs_1.d_max_min()
compl = fobs_1.completeness(d_max = d_max)
n_ref = sel.count(True)
num = flex.sum(flex.abs(f1.data()-f2.data()))
den = flex.sum(flex.abs(f1.data()+f2.data())/2)
r = None
if(den!=0):
r = num/den
cc = flex.linear_correlation(x=f1.data(), y=f2.data()).coefficient()
d_range = fobs_1.binner().bin_legend(
i_bin = i_bin, show_bin_number = False, show_counts = False)
fmt = "%3d: %-17s %4.2f %6d %5.3f %6s"
print >> log, fmt % (i_bin, d_range, compl, n_ref, cc,
format_value("%6.4f", r))
# overall statistics
self.cc = flex.linear_correlation(
x=fobs_1.data(),
y=fobs_2.data()).coefficient()
num = flex.sum(flex.abs(fobs_1.data()-fobs_2.data()))
den = flex.sum(flex.abs(fobs_2.data()+fobs_2.data())/2)
self.r_factor = None
if (den != 0) :
self.r_factor = num / den
# map coefficients
def phase_transfer(miller_array, phase_source):
tmp = miller.array(miller_set = miller_array,
data = flex.double(miller_array.indices().size(), 1)
).phase_transfer(phase_source = phase_source)
return miller.array(miller_set = miller_array,
data = miller_array.data() * tmp.data() )
if (not anomalous) :
diff = miller.array(
miller_set = f_model,
data = fobs_1.data()-fobs_2.data())
self.map_coeff = phase_transfer(
miller_array = diff,
phase_source = f_model)
else :
dano_1 = fobs_1.anomalous_differences()
dano_2 = fobs_2.anomalous_differences()
assert dano_1.indices().all_eq(dano_2.indices())
diff = miller.array(
miller_set = dano_1,
data = dano_1.data() - dano_2.data())
f_model_phases = f_model.average_bijvoet_mates().common_set(diff)
map_coeffs = phase_transfer(
miller_array = diff,
phase_source = f_model_phases)
self.map_coeff = map_coeffs.customized_copy(data=map_coeffs.data()/(2j))
if(self.map_coeff.anomalous_flag()):
self.map_coeff = map_coeff.average_bijvoet_mates()
self.file_names = []
if (write_map) :
self.file_names = self.write_map_file()