def truncate_with_roots(m,
fmodel,
c1,
c2,
cutoff,
scale,
zero_all_interblob_region=True,
as_int=False,
average_peak_volume=None,
selection=None):
assert c1 >= c2
if (average_peak_volume is None):
sites_cart = fmodel.xray_structure.sites_cart()
if (selection is not None):
sites_cart = sites_cart.select(selection)
average_peak_volume = maptbx.peak_volume_estimate(
map_data=m,
sites_cart=sites_cart,
crystal_symmetry=fmodel.xray_structure.crystal_symmetry(),
cutoff=cutoff)
if (average_peak_volume is None
or int(average_peak_volume * scale) - 1 == 0):
return None
average_peak_volume = int(
average_peak_volume * scale /
2) - 1 # XXX "/2" is ad hoc and I don't know why!
co1 = maptbx.connectivity(map_data=m, threshold=c1)
co2 = maptbx.connectivity(map_data=m, threshold=c2)
result = co2.noise_elimination_two_cutoffs(
connectivity_object_at_t1=co1,
elimination_volume_threshold_at_t1=average_peak_volume,
zero_all_interblob_region=zero_all_interblob_region)
if (as_int): return result
else: return result.as_double()
def exercise_expand_mask():
# case 1: standard
cmap = flex.double(flex.grid(30, 30, 30))
cmap.fill(1)
for i in range(10, 20):
for j in range(10, 20):
for k in range(10, 20):
cmap[i, j, k] = 10
co = maptbx.connectivity(map_data=cmap, threshold=5)
new_mask = co.expand_mask(id_to_expand=1, expand_size=1)
for i in range(30):
for j in range(30):
for k in range(30):
assert new_mask[i, j,
k] == (i in range(9, 21) and j in range(9, 21)
and k in range(9, 21))
# case 2: over boundaries
cmap = flex.double(flex.grid(30, 30, 30))
cmap.fill(1)
cmap[1, 1, 1] = 10
co = maptbx.connectivity(map_data=cmap, threshold=5)
new_mask = co.expand_mask(id_to_expand=1, expand_size=2)
for i in range(30):
for j in range(30):
for k in range(30):
assert new_mask[i, j, k] == (i in [29, 0, 1, 2, 3]
and j in [29, 0, 1, 2, 3]
and k in [29, 0, 1, 2, 3])
def exercise_expand_mask():
# case 1: standard
cmap = flex.double(flex.grid(30,30,30))
cmap.fill(1)
for i in range(10,20):
for j in range(10,20):
for k in range(10,20):
cmap[i,j,k] = 10
co = maptbx.connectivity(map_data=cmap, threshold=5)
new_mask = co.expand_mask(id_to_expand=1, expand_size=1)
for i in range(30):
for j in range(30):
for k in range(30):
assert new_mask[i,j,k] == (i in range(9,21) and
j in range(9,21) and k in range(9,21))
# case 2: over boundaries
cmap = flex.double(flex.grid(30,30,30))
cmap.fill(1)
cmap[1,1,1] = 10
co = maptbx.connectivity(map_data=cmap, threshold=5)
new_mask = co.expand_mask(id_to_expand=1, expand_size=2)
for i in range(30):
for j in range(30):
for k in range(30):
assert new_mask[i,j,k] == (i in [29,0,1,2,3] and
j in [29,0,1,2,3] and k in [29,0,1,2,3])
def __init__(self, model, map_data):
adopt_init_args(self, locals())
# Find blob
co = maptbx.connectivity(map_data=self.map_data, threshold=5.)
#connectivity_map = co.result()
#sorted_by_volume = sorted(
# zip(co.regions(), range(0, co.regions().size())), key=lambda x: x[0],
# reverse=True)
#blob_indices = []
#for p in sorted_by_volume:
# v, i = p
# print v, i
# if(i>0):
# blob_indices.append(i)
#######
# You get everything you need:
map_result = co.result()
volumes = co.regions()
print volumes
coors = co.maximum_coors()
vals = co.maximum_values()
minb, maxb = co.get_blobs_boundaries_tuples()
# This will give you the order
i_sorted_by_volume = flex.sort_permutation(
data=volumes, reverse=True) # maybe co.regions() should go there
for i in i_sorted_by_volume:
print "blob #", i
print coors[i]
print vals[i]
print maxb[i], minb[i]
def exercise_sample_all_mask_regions():
cmap = flex.double(flex.grid(30,30,30))
cmap.fill(1)
for i in range(0,10):
for j in range(0,10):
for k in range(0,10):
cmap[i,j,k] = 10
for i in range(15,25):
for j in range(15,25):
for k in range(15,25):
cmap[i,j,k] = 20
co = maptbx.connectivity(map_data=cmap, threshold=5, wrapping=False)
uc = uctbx.unit_cell((10,10,10))
mask_result = co.result()
sample_regs_obj = maptbx.sample_all_mask_regions(
mask=mask_result,
volumes=flex.int([0, 1000,1000]),
sampling_rates=flex.int([0, 10,10]),
unit_cell=uc)
a = sample_regs_obj.get_array(1)
b = sample_regs_obj.get_array(2)
assert a.size() == b.size() == 101
assert approx_equal(a[0], (0,0,0))
assert approx_equal(b[0], (5,5,5))
def prepare_maps(fofc, two_fofc, fem, fofc_cutoff=2, two_fofc_cutoff=0.5,
fem_cutoff=0.5, connectivity_cutoff=0.5, local_average=True):
"""
- This takes 3 maps: mFo-DFc, 2mFo-DFc and FEM and combines them into one map
that is most suitable for real-space refinement.
- Maps are the boxes extracted around region of interest from the whole unit
cell map.
- All maps are expected to be normalized by standard deviation (sigma-scaled)
BEFORE extracting the box. There is no way to assert it at this point.
- Map gridding equivalence is asserted.
"""
m1,m2,m3 = fofc, two_fofc, fem
# assert identical gridding
for m_ in [m1,m2,m3]:
for m__ in [m1,m2,m3]:
assert m_.all() == m__.all()
assert m_.focus() == m__.focus()
assert m_.origin() == m__.origin()
# binarize residual map
sel = m1 <= fofc_cutoff
mask = m1 .set_selected( sel, 0)
mask = mask.set_selected(~sel, 1)
del sel, m1
assert approx_equal([flex.max(mask), flex.min(mask)], [1,0])
def truncate_and_filter(m, cutoff, mask):
return m.set_selected(m<=cutoff, 0)*mask
# truncate and filter 2mFo-DFc map
m2 = truncate_and_filter(m2, two_fofc_cutoff, mask)
# truncate and filter FEM
m3 = truncate_and_filter(m3, fem_cutoff, mask)
del mask
# combined maps
def scale(m):
sd = m.sample_standard_deviation()
if(sd != 0): return m/sd
else: return m
m2 = scale(m2)
m3 = scale(m3)
m = (m2+m3)/2.
del m2, m3
m = scale(m)
# connectivity analysis
co = maptbx.connectivity(map_data=m, threshold=connectivity_cutoff)
v_max=-1.e+9
i_max=None
for i, v in enumerate(co.regions()):
if(i>0):
if(v>v_max):
v_max=v
i_max=i
mask2 = co.result()
selection = mask2==i_max
mask2 = mask2.set_selected(selection, 1)
mask2 = mask2.set_selected(~selection, 0)
assert mask2.count(1) == v_max
# final filter
m = m * mask2.as_double()
if(local_average):
maptbx.map_box_average(map_data=m, cutoff=0.5, index_span=1)
return m
def exercise_volume_cutoff():
cmap = flex.double(flex.grid(100, 100, 100))
cmap.fill(0)
for i in range(100):
for j in range(100):
for k in range(100):
if (5 < i < 10) and (5 < j < 10) and (5 < k < 10):
cmap[i, j, k] = 10
if (15 < i < 25) and (15 < j < 25) and (15 < k < 25):
cmap[i, j, k] = 20
co = maptbx.connectivity(map_data=cmap, threshold=5)
map_result = co.result()
volumes = list(co.regions())
#print volumes
#[999207, 64, 729]
vol_mask = co.volume_cutoff_mask(volume_cutoff=10)
assert (vol_mask == 1).count(True) == 793
assert (vol_mask == 0).count(True) == 999207
vol_mask = co.volume_cutoff_mask(volume_cutoff=100)
assert (vol_mask == 1).count(True) == 729
assert (vol_mask == 0).count(True) == 999271
vol_mask = co.volume_cutoff_mask(volume_cutoff=1000)
assert (vol_mask == 1).count(True) == 0
assert (vol_mask == 0).count(True) == 1000000
def exercise_volume_cutoff():
cmap = flex.double(flex.grid(100,100,100))
cmap.fill(0)
for i in range(100):
for j in range(100):
for k in range(100):
if (5<i<10) and (5<j<10) and (5<k<10):
cmap[i,j,k] = 10
if (15<i<25) and (15<j<25) and (15<k<25):
cmap[i,j,k] = 20
co = maptbx.connectivity(map_data=cmap, threshold=5)
map_result = co.result()
volumes = list(co.regions())
#print volumes
#[999207, 64, 729]
vol_mask = co.volume_cutoff_mask(volume_cutoff=10)
assert (vol_mask==1).count(True) == 793
assert (vol_mask==0).count(True) == 999207
vol_mask = co.volume_cutoff_mask(volume_cutoff=100)
assert (vol_mask==1).count(True) == 729
assert (vol_mask==0).count(True) == 999271
vol_mask = co.volume_cutoff_mask(volume_cutoff=1000)
assert (vol_mask==1).count(True) == 0
assert (vol_mask==0).count(True) == 1000000
def filter_mask(mask_p1,
volume_cutoff,
crystal_symmetry,
for_structure_factors=False):
co = maptbx.connectivity(map_data=mask_p1,
threshold=0.01,
preprocess_against_shallow=True,
wrapping=True)
mi, ma = flex.min(mask_p1), flex.max(mask_p1)
print(mask_p1.size(), (mask_p1 < 0).count(True))
assert mi == 0, mi
assert ma == 1, ma
a, b, c = crystal_symmetry.unit_cell().parameters()[:3]
na, nb, nc = mask_p1.accessor().all()
step = flex.mean(flex.double([a / na, b / nb, c / nc]))
if (crystal_symmetry.space_group_number() != 1):
co.merge_symmetry_related_regions(
space_group=crystal_symmetry.space_group())
conn = co.result().as_double()
z = zip(co.regions(), range(0, co.regions().size()))
sorted_by_volume = sorted(z, key=lambda x: x[0], reverse=True)
for i_seq, p in enumerate(sorted_by_volume):
v, i = p
if (i == 0): continue # skip macromolecule
# skip small volume
volume = v * step**3
if volume < volume_cutoff:
conn = conn.set_selected(conn == i, 0)
conn = conn.set_selected(conn > 0, 1)
if for_structure_factors:
conn = conn / crystal_symmetry.space_group().order_z()
return conn
def exercise_get_blobs_boundaries():
cmap = flex.double(flex.grid(100, 100, 100))
cmap.fill(1)
for i in range(10, 20):
for j in range(10, 20):
for k in range(10, 20):
cmap[i, j, k] = 10
co = maptbx.connectivity(map_data=cmap, threshold=5)
# raw function:
boundaries = co.get_blobs_boundaries()
# how to use this:
# boundaries[min/max, n_blob, x/y/z]
blob_0_min_boundaries = \
(boundaries[0,0,0], boundaries[0,0,1], boundaries[0,0,1])
blob_0_max_boundaries = \
(boundaries[1,0,0], boundaries[1,0,1], boundaries[1,0,1])
# 0th blob - under the limit, covering almost whole cell
assert blob_0_min_boundaries == (0, 0, 0)
assert blob_0_max_boundaries == (99, 99, 99)
# 1st blob - covers coordinates from 10 to 19 by construction
blob_1_min_boundaries = \
(boundaries[0,1,0], boundaries[0,1,1], boundaries[0,1,1])
blob_1_max_boundaries = \
(boundaries[1,1,0], boundaries[1,1,1], boundaries[1,1,1])
assert blob_1_min_boundaries == (10, 10, 10)
assert blob_1_max_boundaries == (19, 19, 19)
# convinient get_blobs_boundaries_tuples
minb, maxb = co.get_blobs_boundaries_tuples()
assert minb == [(0, 0, 0), (10, 10, 10)]
assert maxb == [(99, 99, 99), (19, 19, 19)]
# ==============================
# two blobs test
# just add a blob to the previous cmap
for i in range(50, 70):
for j in range(50, 80):
for k in range(50, 90):
cmap[i, j, k] = 10
co = maptbx.connectivity(map_data=cmap, threshold=5)
minb, maxb = co.get_blobs_boundaries_tuples()
assert minb == [(0, 0, 0), (10, 10, 10), (50, 50, 50)]
assert maxb == [(99, 99, 99), (19, 19, 19), (69, 79, 89)]
def exercise_get_blobs_boundaries():
cmap = flex.double(flex.grid(100,100,100))
cmap.fill(1)
for i in range(10,20):
for j in range(10,20):
for k in range(10,20):
cmap[i,j,k] = 10
co = maptbx.connectivity(map_data=cmap, threshold=5)
# raw function:
boundaries = co.get_blobs_boundaries()
# how to use this:
# boundaries[min/max, n_blob, x/y/z]
blob_0_min_boundaries = \
(boundaries[0,0,0], boundaries[0,0,1], boundaries[0,0,1])
blob_0_max_boundaries = \
(boundaries[1,0,0], boundaries[1,0,1], boundaries[1,0,1])
# 0th blob - under the limit, covering almost whole cell
assert blob_0_min_boundaries == (0,0,0)
assert blob_0_max_boundaries == (99,99,99)
# 1st blob - covers coordinates from 10 to 19 by construction
blob_1_min_boundaries = \
(boundaries[0,1,0], boundaries[0,1,1], boundaries[0,1,1])
blob_1_max_boundaries = \
(boundaries[1,1,0], boundaries[1,1,1], boundaries[1,1,1])
assert blob_1_min_boundaries == (10,10,10)
assert blob_1_max_boundaries == (19,19,19)
# convinient get_blobs_boundaries_tuples
minb, maxb = co.get_blobs_boundaries_tuples()
assert minb == [(0,0,0), (10,10,10)]
assert maxb == [(99,99,99), (19,19,19)]
# ==============================
# two blobs test
# just add a blob to the previous cmap
for i in range(50,70):
for j in range(50,80):
for k in range(50,90):
cmap[i,j,k] = 10
co = maptbx.connectivity(map_data=cmap, threshold=5)
minb, maxb = co.get_blobs_boundaries_tuples()
assert minb == [(0,0,0), (10,10,10), (50,50,50)]
assert maxb == [(99,99,99), (19,19,19), (69,79,89)]
def getvs(cmap, threshold, wrap=True):
co = maptbx.connectivity(map_data=cmap, threshold=threshold, wrapping=wrap)
map_result = co.result()
regs = co.regions()
coors = co.maximum_coors()
vals = co.maximum_values()
assert len(list(regs)) == len(list(coors)) == len(list(vals))
# check dimensions
assert cmap.all() == map_result.all()
v=[0,0,0]
for i in range(3):
v[i] = (map_result==i).count(True)
return v, list(co.regions())
def getvs(cmap, threshold, wrap=True):
co = maptbx.connectivity(map_data=cmap, threshold=threshold, wrapping=wrap)
map_result = co.result()
regs = co.regions()
coors = co.maximum_coors()
vals = co.maximum_values()
assert len(list(regs)) == len(list(coors)) == len(list(vals))
# check dimensions
assert cmap.all() == map_result.all()
v = [0, 0, 0]
for i in range(3):
v[i] = (map_result == i).count(True)
return v, list(co.regions())
def truncate_with_roots(
m, fmodel, c1, c2, cutoff, scale, zero_all_interblob_region=True,
as_int=False, average_peak_volume=None, selection=None):
assert c1>=c2
if(average_peak_volume is None):
sites_cart = fmodel.xray_structure.sites_cart()
if(selection is not None):
sites_cart = sites_cart.select(selection)
average_peak_volume = maptbx.peak_volume_estimate(
map_data = m,
sites_cart = sites_cart,
crystal_symmetry = fmodel.xray_structure.crystal_symmetry(),
cutoff = cutoff)
if(average_peak_volume is None or int(average_peak_volume*scale)-1==0):
return None
average_peak_volume = int(average_peak_volume*scale/2)-1 # XXX "/2" is ad hoc and I don't know why!
co1 = maptbx.connectivity(map_data=m, threshold=c1)
co2 = maptbx.connectivity(map_data=m, threshold=c2)
result = co2.noise_elimination_two_cutoffs(
connectivity_object_at_t1=co1,
elimination_volume_threshold_at_t1=average_peak_volume,
zero_all_interblob_region=zero_all_interblob_region)
if(as_int): return result
else: return result.as_double()
def exercise_max_values():
cmap = flex.double(flex.grid(100, 100, 100))
cmap.fill(0)
for i in range(100):
for j in range(100):
for k in range(100):
if (5 < i < 10) and (5 < j < 10) and (5 < k < 10):
cmap[i, j, k] = 10
if (15 < i < 25) and (15 < j < 25) and (15 < k < 25):
cmap[i, j, k] = 20
cmap[7, 7, 7] = 15
cmap[20, 20, 20] = 25
co = maptbx.connectivity(map_data=cmap, threshold=5)
m_coors = list(co.maximum_coors())
m_vals = list(co.maximum_values())
vols = list(co.regions())
assert len(m_coors) == len(m_vals) == len(vols)
assert m_coors == [(0, 0, 0), (7, 7, 7), (20, 20, 20)]
assert m_vals == [0.0, 15.0, 25.0]
def exercise_max_values():
cmap = flex.double(flex.grid(100,100,100))
cmap.fill(0)
for i in range(100):
for j in range(100):
for k in range(100):
if (5<i<10) and (5<j<10) and (5<k<10):
cmap[i,j,k] = 10
if (15<i<25) and (15<j<25) and (15<k<25):
cmap[i,j,k] = 20
cmap[7,7,7] = 15
cmap[20,20,20] = 25
co = maptbx.connectivity(map_data=cmap, threshold=5)
m_coors = list(co.maximum_coors())
m_vals = list(co.maximum_values())
vols = list(co.regions())
assert len(m_coors) == len(m_vals) == len(vols)
assert m_coors == [(0, 0, 0), (7, 7, 7), (20, 20, 20)]
assert m_vals == [0.0, 15.0, 25.0]
def exercise1():
pdb_str = """
CRYST1 10.000 10.000 10.000 90.00 90.00 90.00 P 1
HETATM 1 C C 1 2.000 2.000 2.000 1.00 20.00 C
END
"""
pdb_inp = iotbx.pdb.input(source_info=None, lines=pdb_str)
xrs = pdb_inp.xray_structure_simple()
cg = maptbx.crystal_gridding(unit_cell=xrs.unit_cell(),
pre_determined_n_real=(100, 100, 100),
space_group_info=xrs.space_group_info())
fc = xrs.structure_factors(d_min=1., algorithm="direct").f_calc()
fft_map = miller.fft_map(crystal_gridding=cg, fourier_coefficients=fc)
map_data = fft_map.real_map_unpadded()
# pass map and threshold value
co = maptbx.connectivity(map_data=map_data, threshold=100.)
# get 'map' of the same size with integers: 0 where below threshold,
# 1,2,3... - for connected regions
map_result = co.result()
# to find out the number of connected region for particular point:
assert map_result[0, 0, 0] == 0 # means under threshold
assert map_result[20, 20, 20] == 1 # blob 1
# get 1d array of integer volumes and transform it to list.
volumes = list(co.regions())
# find max volume (except volume of 0-region which will be probably max)
max_volume = max(volumes[1:])
# find number of the region with max volume
max_index = volumes.index(max_volume)
v = [0, 0, 0]
for i in range(3):
# !!! Do not do this because it's extremely slow! Used for test purposes.
v[i] = (map_result == i).count(True)
assert v[2] == 0
assert v[1] < 15000
assert v[0] + v[1] + v[2] == 1000000
assert volumes == v[:2]
def exercise1():
pdb_str="""
CRYST1 10.000 10.000 10.000 90.00 90.00 90.00 P 1
HETATM 1 C C 1 2.000 2.000 2.000 1.00 20.00 C
END
"""
pdb_inp = iotbx.pdb.input(source_info=None, lines=pdb_str)
xrs = pdb_inp.xray_structure_simple()
cg = maptbx.crystal_gridding(unit_cell=xrs.unit_cell(),
pre_determined_n_real=(100,100,100),
space_group_info=xrs.space_group_info())
fc = xrs.structure_factors(d_min = 1., algorithm = "direct").f_calc()
fft_map = miller.fft_map(crystal_gridding=cg, fourier_coefficients=fc)
map_data = fft_map.real_map_unpadded()
# pass map and threshold value
co = maptbx.connectivity(map_data=map_data, threshold=100.)
# get 'map' of the same size with integers: 0 where below threshold,
# 1,2,3... - for connected regions
map_result = co.result()
# to find out the number of connected region for particular point:
assert map_result[0,0,0] == 0 # means under threshold
assert map_result[20,20,20] == 1 # blob 1
# get 1d array of integer volumes and transform it to list.
volumes = list(co.regions())
# find max volume (except volume of 0-region which will be probably max)
max_volume = max(volumes[1:])
# find number of the region with max volume
max_index = volumes.index(max_volume)
v=[0,0,0]
for i in range(3):
# !!! Do not do this because it's extremely slow! Used for test purposes.
v[i] = (map_result==i).count(True)
assert v[2] == 0
assert v[1] < 15000
assert v[0]+v[1]+v[2] == 1000000
assert volumes == v[:2]
def exercise_wrapping():
cmap = flex.double(flex.grid(30, 30, 30))
cmap.fill(1)
for i in range(0, 5):
for j in range(0, 5):
for k in range(0, 5):
cmap[i, j, k] = 10
for i in range(0, 5):
for j in range(25, 30):
for k in range(0, 5):
cmap[i, j, k] = 10
for i in range(0, 5):
for j in range(0, 5):
for k in range(25, 30):
cmap[i, j, k] = 10
for i in range(25, 30):
for j in range(0, 5):
for k in range(0, 5):
cmap[i, j, k] = 10
for i in range(25, 30):
for j in range(25, 30):
for k in range(0, 5):
cmap[i, j, k] = 10
for i in range(25, 30):
for j in range(0, 5):
for k in range(25, 30):
cmap[i, j, k] = 10
n_in_blob = cmap.count(10)
co = maptbx.connectivity(map_data=cmap, threshold=5, wrapping=True)
dres = co.result().as_double()
regs = list(co.regions())
minb, maxb = co.get_blobs_boundaries_tuples()
assert n_in_blob == 750
assert regs == [26250, 750]
def __init__(
self,
xray_structure,
step,
volume_cutoff=None,
mean_diff_map_threshold=None,
compute_whole=False,
largest_only=False,
wrapping=True, # should be False if working with ASU
f_obs=None,
r_sol=1.1,
r_shrink=0.9,
f_calc=None,
log=None,
write_masks=False):
adopt_init_args(self, locals())
#
self.d_spacings = f_obs.d_spacings().data()
self.sel_gte3 = self.d_spacings >= 3
self.miller_array = f_obs.select(self.sel_gte3)
#
self.crystal_symmetry = self.xray_structure.crystal_symmetry()
# Compute mask in p1 (via ASU)
self.crystal_gridding = maptbx.crystal_gridding(
unit_cell=xray_structure.unit_cell(),
space_group_info=xray_structure.space_group_info(),
symmetry_flags=maptbx.use_space_group_symmetry,
step=step)
self.n_real = self.crystal_gridding.n_real()
# XXX Where do we want to deal with H and occ==0?
self._mask_p1 = self._compute_mask_in_p1()
self.solvent_content = 100.*(self._mask_p1 != 0).count(True)/\
self._mask_p1.size()
# Optionally compute Fmask from original whole mask, zero-ed at dmin<3A.
self.f_mask_whole = self._compute_f_mask_whole()
# Connectivity analysis
co = maptbx.connectivity(map_data=self._mask_p1,
threshold=0.01,
preprocess_against_shallow=False,
wrapping=wrapping)
if (xray_structure.space_group().type().number() != 1): # not P1
co.merge_symmetry_related_regions(
space_group=xray_structure.space_group())
#
self.conn = co.result().as_double()
z = zip(co.regions(), range(0, co.regions().size()))
sorted_by_volume = sorted(z, key=lambda x: x[0], reverse=True)
#
f_mask_data_0 = flex.complex_double(f_obs.data().size(), 0)
self.f_mask_0 = None
self.FV = OrderedDict()
self.mFoDFc_0 = None
diff_map = None # mFo-DFc map computed using F_mask_0 (main mask)
self.regions = OrderedDict()
small_selection = None
weak_selection = None
#
if (log is not None):
print(" # volume_p1 uc(%) mFo-DFc: min,max,mean,sd",
file=log)
#
for i_seq, p in enumerate(sorted_by_volume):
v, i = p
self._region_i_selection = None # must be here inside the loop!
f_mask_i = None # must be here inside the loop!
# skip macromolecule
if (i == 0): continue
# skip small volume and accumulate small volumes
volume = v * step**3
uc_fraction = v * 100. / self.conn.size()
if (volume_cutoff is not None and volume < volume_cutoff):
if (volume >= 10):
if (small_selection is None):
small_selection = self._get_region_i_selection(i)
else:
small_selection |= self._get_region_i_selection(i)
continue
# Accumulate regions with volume greater than volume_cutoff (if
# volume_cutoff is defined). Weak density regions are included.
self.regions[i_seq] = group_args(id=i,
i_seq=i_seq,
volume=volume,
uc_fraction=uc_fraction)
# Compute i-th region mask
mask_i_asu = self.compute_i_mask_asu(
selection=self._get_region_i_selection(i), volume=volume)
# Compute F_mask_0 (F_mask for main mask)
if (uc_fraction >= 1):
f_mask_i = self.compute_f_mask_i(mask_i_asu)
f_mask_data_0 += f_mask_i.data()
elif (largest_only):
break
# Compute mFo-DFc map using main mask (once done computing main mask!)
if (uc_fraction < 1 and diff_map is None):
diff_map = self.compute_diff_map(f_mask_data_0=f_mask_data_0)
# Analyze mFo-DFc map in the i-th region
mi, ma, me, sd = None, None, None, None
if (diff_map is not None):
iselection = self._get_region_i_selection(i).iselection()
blob = diff_map.select(iselection)
mean_diff_map = flex.mean(diff_map.select(iselection))
mi, ma, me = flex.min(blob), flex.max(blob), flex.mean(blob)
sd = blob.sample_standard_deviation()
if (log is not None):
print("%3d" % i_seq,
"%12.3f" % volume,
"%8.4f" % round(uc_fraction, 4),
"%7.3f %7.3f %7.3f %7.3f" % (mi, ma, me, sd),
file=log)
# Accumulate regions with weak density into one region, then skip
if (mean_diff_map_threshold is not None):
if (mean_diff_map <= mean_diff_map_threshold):
if (mean_diff_map > 0.1):
if (weak_selection is None):
weak_selection = self._get_region_i_selection(
i)
else:
weak_selection |= self._get_region_i_selection(
i)
continue
else:
if (log is not None):
print("%3d" % i_seq,
"%12.3f" % volume,
"%8.4f" % round(uc_fraction, 4),
"%7s" % str(None),
file=log)
# Compute F_maks for i-th region
if (f_mask_i is None):
f_mask_i = self.compute_f_mask_i(mask_i_asu)
# Compose result object
self.FV[f_mask_i] = [round(volume, 3), round(uc_fraction, 1)]
#
# Determine number of secondary regions. Must happen here!
# Preliminarily if need to do mosaic.
self.n_regions = len(self.FV.values())
self.do_mosaic = False
if (self.n_regions > 1 and flex.max(self.d_spacings) > 6):
self.do_mosaic = True
# Add aggregated small regions (if present)
self._add_from_aggregated(selection=small_selection, diff_map=diff_map)
# Add aggregated weak map regions (if present)
self._add_from_aggregated(selection=weak_selection, diff_map=diff_map)
# Finalize main Fmask
self.f_mask_0 = f_obs.customized_copy(data=f_mask_data_0)
# Delete bulk whole mask from memory
del self._mask_p1
def __init__(self, fmodel, log=None):
# Commonly used objects
xrs = fmodel.xray_structure
sgt = xrs.space_group().type()
# Compute default fmodel and decide on grid step
fmodel, self.grid_step_factor = get_fmodel_and_grid_step(
f_obs = fmodel.f_obs(),
r_free_flags = fmodel.r_free_flags(),
xrs = xrs)
#fmodel.show()
#print fmodel.r_work(), fmodel.r_free()
###
mask_data_p1, n_real, crystal_gridding = get_mask_1(fmodel=fmodel,
grid_step_factor=self.grid_step_factor)
#ccp4_map(cg=crystal_gridding, file_name="m1.ccp4", map_data=mask_data_p1_)
#xxx1 = fmodel.f_obs().structure_factors_from_map(map=mask_data_p1,
# use_scale = True, anomalous_flag = False, use_sg = False)
#mask_data_p1, n_real, crystal_gridding = get_mask_2(fmodel=fmodel,
# grid_step_factor=self.grid_step_factor)
#print n_real
#STOP()
#xxx2 = fmodel.f_obs().structure_factors_from_map(map=mask_data_p1,
# use_scale = True, anomalous_flag = False, use_sg = False)
#
#assert approx_equal(xxx1.data(), xxx2.data())
#print mask_data_p1.all(), mask_data_p1.focus(), mask_data_p1.origin()
#print mask_data_p1_2.all(), mask_data_p1_2.focus(), mask_data_p1_2.origin()
#print mask_data_p1_1.count(0), mask_data_p1_2.count(0)
#assert approx_equal(mask_data_p1_1, mask_data_p1_2)
#STOP()
#####
# Mask connectivity analysis
co = maptbx.connectivity(map_data=mask_data_p1, threshold=0.01)
conn = co.result().as_double()
# Convert result of connectivity analysis from P1 to ASU (in-place)
conn = asu_map_ext.asymmetric_map(sgt, conn).data()
# Find unique indices and regions in reduced (P1->ASU) conn
region_indices = flex.double()
region_volumes = flex.double()
for i in conn:
if not i in region_indices: region_indices.append(i)
for l in region_indices:
szl = conn.count(l)*100./conn.size()
region_volumes.append(szl)
s = flex.sort_permutation(region_volumes, reverse=True)
region_volumes = region_volumes.select(s)
region_indices = region_indices.select(s)
# Convert P1 mask into ASU
mask_data_asu = asu_map_ext.asymmetric_map(sgt, mask_data_p1).data()
conn.reshape(mask_data_asu.accessor()) #XXX still need it?
f_masks = []
all_zero_found = False
if(log is not None): print >> log, "Number of regions:", len(region_indices)
mi,ma,me,diff_map_asu = None,None,None,None
for ii, i in enumerate(region_indices):
s = conn==i
si = s.iselection()
if(not all_zero_found and mask_data_asu.select(si).count(0.)>0):
all_zero_found = True
continue
# DIFF MAP START
if(region_volumes[ii]<1 and diff_map_asu is None):#(ii == 2):
fmodel_tmp = mmtbx.f_model.manager(
f_obs = fmodel.f_obs(),
r_free_flags = fmodel.r_free_flags(),
f_calc = fmodel.f_calc(),
f_mask = f_masks[len(f_masks)-1])
fmodel_tmp.update_all_scales(remove_outliers=False, update_f_part1=False)
diff_map_p1 = compute_map(
fmodel = fmodel_tmp,
crystal_gridding = crystal_gridding,
map_type = "mFo-DFc")
diff_map_asu = asu_map_ext.asymmetric_map(sgt, diff_map_p1).data()
if(diff_map_asu is not None):
mi,ma,me = diff_map_asu.select(si).min_max_mean().as_tuple()
if(ma<0. or me<0.):
continue
# DIFF MAP END
#XXX this is 4 loops, may be slow. move to C++ if slow.
mask_data_asu_i = mask_data_asu.deep_copy()
#mask_data_asu_i = mask_data_asu_i.set_selected(s, 1).set_selected(~s, 0)
mask_data_asu_i = mask_data_asu_i.set_selected(~s, 0)
#if(mi is None):
# print "region: %5d fraction: %8.4f"%(ii, region_volumes[ii]), len(region_volumes)
#else:
# print "region: %5d fraction: %8.4f"%(ii, region_volumes[ii]), len(region_volumes), "%7.3f %7.3f %7.3f"%(mi,ma,me)
if(log is not None):
print >> log, "region: %5d fraction: %8.4f"%(ii, region_volumes[ii])
log.flush()
f_mask_i = fmodel.f_obs().structure_factors_from_asu_map(
asu_map_data = mask_data_asu_i, n_real = n_real)
if(len(f_masks)>0 and region_volumes[ii]>1):
f_masks[len(f_masks)-1] = f_masks[len(f_masks)-1].array(data = f_masks[len(f_masks)-1].data()+
f_mask_i.data())
else:
f_masks.append(f_mask_i)
#
self.fmodel_result, self.method = helper_3(
fmodel = fmodel,
f_masks = f_masks,
log = log)
#self.fmodel_result.show()
#
self.n_regions = len(region_volumes[1:])
self.region_volumes = " ".join(["%8.4f"%(v) for v in region_volumes[1:][:10]]) # top 10
def exercise_noise_elimination_two_cutoffs():
# Purpose: eliminate noise.
# We want to delete small blobs from the map. On the particular contouring
# (cutoff) level we can set a threshold for volume and say: all blobs that
# have volume less than threshold value should be deleted.
# One more point is that we want to delete them with their 'root', meaning
# that we are lowering threshold level and put zeros on that bigger regions.
# But we are zeroing only those which are not merged with big good blobs.
# Everything under second contouring level also will be zero.
# ======================
# From another point of view.
# We know some threshold value for volume of good blobs on t1 contouring
# level. We want to keep only them and clear out everything else. But the
# keeping and clearing should be done at lower t2 contouring level.
#
# The result (res_mask) is 3d integer array sized as original map.
# res_mask contain 0 for noise, 1 for valuable information.
# Mask corresponding to t2 contouring level.
#
# The option "zero_all_interblob_region" by default is True, and this means
# that everything below threshold on t2 level will be 0. If
# zero_all_interblob_region=False then everything below threshold on t2
# level will be 1.
#
#map preparation for test
cmap = flex.double(flex.grid(100,2,2))
cmap.fill(10)
for i in range(10,40):
cmap[i,1,1] = i
for i,v in zip(range(40,60), range(40,20,-1)):
cmap[i,1,1] = v
for i,v in zip(range(60,70), range(20,30)):
cmap[i,1,1] = v
for i,v in zip(range(70,90), range(30,10,-1)):
cmap[i,1,1] = v
#for i in range(100):
# print "%d : %d" % (i, cmap[i,1,1])
co1 = maptbx.connectivity(map_data=cmap, threshold=25)
co2 = maptbx.connectivity(map_data=cmap, threshold=22)
co3 = maptbx.connectivity(map_data=cmap, threshold=18)
# Example 1. We have one good blob (volume>12) and one bad (volume < 12).
# After lowering contour level they are still separate, so we want to keep
# only big first blob, which has volume=35 on t2 contour level.
# Here is actual call to get a mask.
res_mask = co2.noise_elimination_two_cutoffs(
connectivity_object_at_t1=co1,
elimination_volume_threshold_at_t1=12,
zero_all_interblob_region=True)
assert (res_mask!=0).count(True) == 35
# 2 good ===> 2 separate
res_mask = co2.noise_elimination_two_cutoffs(
connectivity_object_at_t1=co1,
elimination_volume_threshold_at_t1=8)
assert (res_mask!=0).count(True) == 50
# 1 good, 1 bad ===> 1 big
res_mask = co3.noise_elimination_two_cutoffs(
connectivity_object_at_t1=co1,
elimination_volume_threshold_at_t1=12)
assert (res_mask!=0).count(True) == 63
# 2 good ===> 1 big
res_mask = co3.noise_elimination_two_cutoffs(
connectivity_object_at_t1=co1,
elimination_volume_threshold_at_t1=8)
assert (res_mask!=0).count(True) == 63
# 2 bad ===> 1 big
res_mask = co3.noise_elimination_two_cutoffs(
connectivity_object_at_t1=co1,
elimination_volume_threshold_at_t1=30)
assert (res_mask!=0).count(True) == 0
# extreme case: nothing above t1 ==> result: everything is 0 on the mask
co1 = maptbx.connectivity(map_data=cmap, threshold=40)
co2 = maptbx.connectivity(map_data=cmap, threshold=22)
res_mask = co2.noise_elimination_two_cutoffs(
connectivity_object_at_t1=co1,
elimination_volume_threshold_at_t1=10)
assert (res_mask!=0).count(True) == 0
# extreme case: everything above t1 ==> result is undefined.
# =================================================================
# same as above, but zero_all_interblob_region = False
# In the first test we have 1 good blob and one bad blob. Bad one
# will have volume=15 on t2 contouring level so we want to have 385 non-zeros
# on resulting mask
co1 = maptbx.connectivity(map_data=cmap, threshold=25)
co2 = maptbx.connectivity(map_data=cmap, threshold=22)
co3 = maptbx.connectivity(map_data=cmap, threshold=18)
res_mask = co2.noise_elimination_two_cutoffs(
connectivity_object_at_t1=co1,
elimination_volume_threshold_at_t1=12,
zero_all_interblob_region=False)
#for i in range(100):
# print "%d : %d | %d" % (i, cmap[i,1,1], res_mask[i,1,1])
assert (res_mask!=0).count(True) == 385
# 2 good ===> 2 separate
res_mask = co2.noise_elimination_two_cutoffs(
connectivity_object_at_t1=co1,
elimination_volume_threshold_at_t1=8,
zero_all_interblob_region=False)
assert (res_mask==1).count(True) == 400
# 1 good, 1 bad ===> 1 big
res_mask = co3.noise_elimination_two_cutoffs(
connectivity_object_at_t1=co1,
elimination_volume_threshold_at_t1=12,
zero_all_interblob_region=False)
assert (res_mask!=0).count(True) == 400
# 2 good ===> 1 big
res_mask = co3.noise_elimination_two_cutoffs(
connectivity_object_at_t1=co1,
elimination_volume_threshold_at_t1=8,
zero_all_interblob_region=False)
assert (res_mask!=0).count(True) == 400
# 2 bad ===> 1 big
res_mask = co3.noise_elimination_two_cutoffs(
connectivity_object_at_t1=co1,
elimination_volume_threshold_at_t1=30,
zero_all_interblob_region=False)
assert (res_mask!=0).count(True) == 337
# extreme case: nothing above t1, something above t2 ==> result:
# everything between blobs on t2 will be 1.
co1 = maptbx.connectivity(map_data=cmap, threshold=40)
co2 = maptbx.connectivity(map_data=cmap, threshold=22)
res_mask = co2.noise_elimination_two_cutoffs(
connectivity_object_at_t1=co1,
elimination_volume_threshold_at_t1=10,
zero_all_interblob_region=False)
assert (res_mask!=0).count(True) == 350
def exercise_noise_elimination_two_cutoffs():
# Purpose: eliminate noise.
# We want to delete small blobs from the map. On the particular contouring
# (cutoff) level we can set a threshold for volume and say: all blobs that
# have volume less than threshold value should be deleted.
# One more point is that we want to delete them with their 'root', meaning
# that we are lowering threshold level and put zeros on that bigger regions.
# But we are zeroing only those which are not merged with big good blobs.
# Everything under second contouring level also will be zero.
# ======================
# From another point of view.
# We know some threshold value for volume of good blobs on t1 contouring
# level. We want to keep only them and clear out everything else. But the
# keeping and clearing should be done at lower t2 contouring level.
#
# The result (res_mask) is 3d integer array sized as original map.
# res_mask contain 0 for noise, 1 for valuable information.
# Mask corresponding to t2 contouring level.
#
# The option "zero_all_interblob_region" by default is True, and this means
# that everything below threshold on t2 level will be 0. If
# zero_all_interblob_region=False then everything below threshold on t2
# level will be 1.
#
#map preparation for test
cmap = flex.double(flex.grid(100, 2, 2))
cmap.fill(10)
for i in range(10, 40):
cmap[i, 1, 1] = i
for i, v in zip(range(40, 60), range(40, 20, -1)):
cmap[i, 1, 1] = v
for i, v in zip(range(60, 70), range(20, 30)):
cmap[i, 1, 1] = v
for i, v in zip(range(70, 90), range(30, 10, -1)):
cmap[i, 1, 1] = v
#for i in range(100):
# print "%d : %d" % (i, cmap[i,1,1])
co1 = maptbx.connectivity(map_data=cmap, threshold=25)
co2 = maptbx.connectivity(map_data=cmap, threshold=22)
co3 = maptbx.connectivity(map_data=cmap, threshold=18)
# Example 1. We have one good blob (volume>12) and one bad (volume < 12).
# After lowering contour level they are still separate, so we want to keep
# only big first blob, which has volume=35 on t2 contour level.
# Here is actual call to get a mask.
res_mask = co2.noise_elimination_two_cutoffs(
connectivity_object_at_t1=co1,
elimination_volume_threshold_at_t1=12,
zero_all_interblob_region=True)
assert (res_mask != 0).count(True) == 35
# 2 good ===> 2 separate
res_mask = co2.noise_elimination_two_cutoffs(
connectivity_object_at_t1=co1, elimination_volume_threshold_at_t1=8)
assert (res_mask != 0).count(True) == 50
# 1 good, 1 bad ===> 1 big
res_mask = co3.noise_elimination_two_cutoffs(
connectivity_object_at_t1=co1, elimination_volume_threshold_at_t1=12)
assert (res_mask != 0).count(True) == 63
# 2 good ===> 1 big
res_mask = co3.noise_elimination_two_cutoffs(
connectivity_object_at_t1=co1, elimination_volume_threshold_at_t1=8)
assert (res_mask != 0).count(True) == 63
# 2 bad ===> 1 big
res_mask = co3.noise_elimination_two_cutoffs(
connectivity_object_at_t1=co1, elimination_volume_threshold_at_t1=30)
assert (res_mask != 0).count(True) == 0
# extreme case: nothing above t1 ==> result: everything is 0 on the mask
co1 = maptbx.connectivity(map_data=cmap, threshold=40)
co2 = maptbx.connectivity(map_data=cmap, threshold=22)
res_mask = co2.noise_elimination_two_cutoffs(
connectivity_object_at_t1=co1, elimination_volume_threshold_at_t1=10)
assert (res_mask != 0).count(True) == 0
# extreme case: everything above t1 ==> result is undefined.
# =================================================================
# same as above, but zero_all_interblob_region = False
# In the first test we have 1 good blob and one bad blob. Bad one
# will have volume=15 on t2 contouring level so we want to have 385 non-zeros
# on resulting mask
co1 = maptbx.connectivity(map_data=cmap, threshold=25)
co2 = maptbx.connectivity(map_data=cmap, threshold=22)
co3 = maptbx.connectivity(map_data=cmap, threshold=18)
res_mask = co2.noise_elimination_two_cutoffs(
connectivity_object_at_t1=co1,
elimination_volume_threshold_at_t1=12,
zero_all_interblob_region=False)
#for i in range(100):
# print "%d : %d | %d" % (i, cmap[i,1,1], res_mask[i,1,1])
assert (res_mask != 0).count(True) == 385
# 2 good ===> 2 separate
res_mask = co2.noise_elimination_two_cutoffs(
connectivity_object_at_t1=co1,
elimination_volume_threshold_at_t1=8,
zero_all_interblob_region=False)
assert (res_mask == 1).count(True) == 400
# 1 good, 1 bad ===> 1 big
res_mask = co3.noise_elimination_two_cutoffs(
connectivity_object_at_t1=co1,
elimination_volume_threshold_at_t1=12,
zero_all_interblob_region=False)
assert (res_mask != 0).count(True) == 400
# 2 good ===> 1 big
res_mask = co3.noise_elimination_two_cutoffs(
connectivity_object_at_t1=co1,
elimination_volume_threshold_at_t1=8,
zero_all_interblob_region=False)
assert (res_mask != 0).count(True) == 400
# 2 bad ===> 1 big
res_mask = co3.noise_elimination_two_cutoffs(
connectivity_object_at_t1=co1,
elimination_volume_threshold_at_t1=30,
zero_all_interblob_region=False)
assert (res_mask != 0).count(True) == 337
# extreme case: nothing above t1, something above t2 ==> result:
# everything between blobs on t2 will be 1.
co1 = maptbx.connectivity(map_data=cmap, threshold=40)
co2 = maptbx.connectivity(map_data=cmap, threshold=22)
res_mask = co2.noise_elimination_two_cutoffs(
connectivity_object_at_t1=co1,
elimination_volume_threshold_at_t1=10,
zero_all_interblob_region=False)
assert (res_mask != 0).count(True) == 350
def __init__(self, fmodel, log=None):
# Commonly used objects
xrs = fmodel.xray_structure
sgt = xrs.space_group().type()
# Compute default fmodel and decide on grid step
fmodel, self.grid_step_factor = get_fmodel_and_grid_step(
f_obs=fmodel.f_obs(), r_free_flags=fmodel.r_free_flags(), xrs=xrs)
#fmodel.show()
#print fmodel.r_work(), fmodel.r_free()
###
mask_data_p1, n_real, crystal_gridding = get_mask_1(
fmodel=fmodel, grid_step_factor=self.grid_step_factor)
#ccp4_map(cg=crystal_gridding, file_name="m1.ccp4", map_data=mask_data_p1_)
#xxx1 = fmodel.f_obs().structure_factors_from_map(map=mask_data_p1,
# use_scale = True, anomalous_flag = False, use_sg = False)
#mask_data_p1, n_real, crystal_gridding = get_mask_2(fmodel=fmodel,
# grid_step_factor=self.grid_step_factor)
#print n_real
#STOP()
#xxx2 = fmodel.f_obs().structure_factors_from_map(map=mask_data_p1,
# use_scale = True, anomalous_flag = False, use_sg = False)
#
#assert approx_equal(xxx1.data(), xxx2.data())
#print mask_data_p1.all(), mask_data_p1.focus(), mask_data_p1.origin()
#print mask_data_p1_2.all(), mask_data_p1_2.focus(), mask_data_p1_2.origin()
#print mask_data_p1_1.count(0), mask_data_p1_2.count(0)
#assert approx_equal(mask_data_p1_1, mask_data_p1_2)
#STOP()
#####
# Mask connectivity analysis
co = maptbx.connectivity(map_data=mask_data_p1, threshold=0.01)
conn = co.result().as_double()
# Convert result of connectivity analysis from P1 to ASU (in-place)
conn = asu_map_ext.asymmetric_map(sgt, conn).data()
# Find unique indices and regions in reduced (P1->ASU) conn
region_indices = flex.double()
region_volumes = flex.double()
for i in conn:
if not i in region_indices: region_indices.append(i)
for l in region_indices:
szl = conn.count(l) * 100. / conn.size()
region_volumes.append(szl)
s = flex.sort_permutation(region_volumes, reverse=True)
region_volumes = region_volumes.select(s)
region_indices = region_indices.select(s)
# Convert P1 mask into ASU
mask_data_asu = asu_map_ext.asymmetric_map(sgt, mask_data_p1).data()
conn.reshape(mask_data_asu.accessor()) #XXX still need it?
f_masks = []
all_zero_found = False
if (log is not None):
print("Number of regions:", len(region_indices), file=log)
mi, ma, me, diff_map_asu = None, None, None, None
for ii, i in enumerate(region_indices):
s = conn == i
si = s.iselection()
if (not all_zero_found and mask_data_asu.select(si).count(0.) > 0):
all_zero_found = True
continue
# DIFF MAP START
if (region_volumes[ii] < 1 and diff_map_asu is None): #(ii == 2):
fmodel_tmp = mmtbx.f_model.manager(
f_obs=fmodel.f_obs(),
r_free_flags=fmodel.r_free_flags(),
f_calc=fmodel.f_calc(),
f_mask=f_masks[len(f_masks) - 1])
fmodel_tmp.update_all_scales(remove_outliers=False,
update_f_part1=False)
diff_map_p1 = compute_map(fmodel=fmodel_tmp,
crystal_gridding=crystal_gridding,
map_type="mFo-DFc")
diff_map_asu = asu_map_ext.asymmetric_map(sgt,
diff_map_p1).data()
if (diff_map_asu is not None):
mi, ma, me = diff_map_asu.select(si).min_max_mean().as_tuple()
if (ma < 0. or me < 0.):
continue
# DIFF MAP END
#XXX this is 4 loops, may be slow. move to C++ if slow.
mask_data_asu_i = mask_data_asu.deep_copy()
#mask_data_asu_i = mask_data_asu_i.set_selected(s, 1).set_selected(~s, 0)
mask_data_asu_i = mask_data_asu_i.set_selected(~s, 0)
#if(mi is None):
# print "region: %5d fraction: %8.4f"%(ii, region_volumes[ii]), len(region_volumes)
#else:
# print "region: %5d fraction: %8.4f"%(ii, region_volumes[ii]), len(region_volumes), "%7.3f %7.3f %7.3f"%(mi,ma,me)
if (log is not None):
print("region: %5d fraction: %8.4f" % (ii, region_volumes[ii]),
file=log)
log.flush()
f_mask_i = fmodel.f_obs().structure_factors_from_asu_map(
asu_map_data=mask_data_asu_i, n_real=n_real)
if (len(f_masks) > 0 and region_volumes[ii] > 1):
f_masks[len(f_masks) - 1] = f_masks[len(f_masks) - 1].array(
data=f_masks[len(f_masks) - 1].data() + f_mask_i.data())
else:
f_masks.append(f_mask_i)
#
self.fmodel_result, self.method = helper_3(fmodel=fmodel,
f_masks=f_masks,
log=log)
#self.fmodel_result.show()
#
self.n_regions = len(region_volumes[1:])
self.region_volumes = " ".join(
["%8.4f" % (v) for v in region_volumes[1:][:10]]) # top 10
def remove_model_density(map_data, xrs, rad_inside=2):
#
map_data = map_data - flex.mean(map_data)
map_data = map_data.set_selected(map_data < 0, 0)
sd = map_data.sample_standard_deviation()
assert sd != 0
map_data = map_data / sd
#
map_at_atoms = flex.double()
for site_frac in xrs.sites_frac():
mv = map_data.tricubic_interpolation(site_frac)
map_at_atoms.append( mv )
print (flex.mean(map_at_atoms), flex.max(map_at_atoms))
mmax = flex.max(map_at_atoms)
cut = 0
print (dir(map_data))
while cut<mmax:
map_data_ = map_data.deep_copy()
map_data_ = map_data_.set_selected(map_data<cut, 0)
map_data_ = map_data_.set_selected(map_data>=cut, 1)
cut+=1
zz = flex.double()
for site_frac in xrs.sites_frac():
mv = map_data_.value_at_closest_grid_point(site_frac)
zz.append( mv )
print(cut, (zz==1).count(True)/zz.size()*100. )
#
#radii = flex.double(xrs.sites_frac().size(), rad_inside)
#mask = cctbx_maptbx_ext.mask(
# sites_frac = xrs.sites_frac(),
# unit_cell = xrs.unit_cell(),
# n_real = map_data.all(),
# mask_value_inside_molecule = 0,
# mask_value_outside_molecule = 1,
# radii = radii)
mask = mmtbx.masks.mask_from_xray_structure(
xray_structure = xrs,
p1 = True,
for_structure_factors = True,
solvent_radius = None,
shrink_truncation_radius = None,
n_real = map_data.accessor().all(),
in_asu = False).mask_data
maptbx.unpad_in_place(map=mask)
map_data = map_data * mask
map_data = map_data.set_selected(map_data < flex.mean(map_at_atoms)/6, 0)
#
n = map_data.accessor().all()
abc = xrs.unit_cell().parameters()[:3]
print(abc[0]/n[0], abc[1]/n[1], abc[2]/n[2])
step = abc[0]/n[0]
co = maptbx.connectivity(
map_data = map_data.deep_copy(),
threshold = 0.0,
preprocess_against_shallow = True,
wrapping = False)
conn = co.result().as_double()
z = zip(co.regions(),range(0,co.regions().size()))
sorted_by_volume = sorted(z, key=lambda x: x[0], reverse=True)
mask_ = flex.double(flex.grid(n), 0)
for i_seq, p in enumerate(sorted_by_volume):
v, i = p
if i_seq==0: continue
volume = v*step**3
print(v, volume)
if 1:#(volume<3):
sel = conn==i
mask_ = mask_.set_selected(sel, 1)
#
return map_data*mask_
def prepare_maps(fofc,
two_fofc,
fem,
fofc_cutoff=2,
two_fofc_cutoff=0.5,
fem_cutoff=0.5,
connectivity_cutoff=0.5,
local_average=True):
"""
- This takes 3 maps: mFo-DFc, 2mFo-DFc and FEM and combines them into one map
that is most suitable for real-space refinement.
- Maps are the boxes extracted around region of interest from the whole unit
cell map.
- All maps are expected to be normalized by standard deviation (sigma-scaled)
BEFORE extracting the box. There is no way to assert it at this point.
- Map gridding equivalence is asserted.
"""
m1, m2, m3 = fofc, two_fofc, fem
# assert identical gridding
for m_ in [m1, m2, m3]:
for m__ in [m1, m2, m3]:
assert m_.all() == m__.all()
assert m_.focus() == m__.focus()
assert m_.origin() == m__.origin()
# binarize residual map
sel = m1 <= fofc_cutoff
mask = m1.set_selected(sel, 0)
mask = mask.set_selected(~sel, 1)
del sel, m1
assert approx_equal([flex.max(mask), flex.min(mask)], [1, 0])
def truncate_and_filter(m, cutoff, mask):
return m.set_selected(m <= cutoff, 0) * mask
# truncate and filter 2mFo-DFc map
m2 = truncate_and_filter(m2, two_fofc_cutoff, mask)
# truncate and filter FEM
m3 = truncate_and_filter(m3, fem_cutoff, mask)
del mask
# combined maps
def scale(m):
sd = m.sample_standard_deviation()
if (sd != 0): return m / sd
else: return m
m2 = scale(m2)
m3 = scale(m3)
m = (m2 + m3) / 2.
del m2, m3
m = scale(m)
# connectivity analysis
co = maptbx.connectivity(map_data=m, threshold=connectivity_cutoff)
v_max = -1.e+9
i_max = None
for i, v in enumerate(co.regions()):
if (i > 0):
if (v > v_max):
v_max = v
i_max = i
mask2 = co.result()
selection = mask2 == i_max
mask2 = mask2.set_selected(selection, 1)
mask2 = mask2.set_selected(~selection, 0)
assert mask2.count(1) == v_max
# final filter
m = m * mask2.as_double()
if (local_average):
maptbx.map_box_average(map_data=m, cutoff=0.5, index_span=1)
return m
def exercise_work_in_asu():
pdb_str = """
CRYST1 10.000 10.000 10.000 90.00 90.00 90.00 P 4
HETATM 1 C C 1 2.000 2.000 2.000 1.00 20.00 C
HETATM 2 C C 2 4.000 4.000 4.000 1.00 20.00 C
END
"""
from time import time
pdb_inp = iotbx.pdb.input(source_info=None, lines=pdb_str)
xrs = pdb_inp.xray_structure_simple()
# xrs.show_summary()
d_min = 1
fc = xrs.structure_factors(d_min=d_min).f_calc()
symmetry_flags = maptbx.use_space_group_symmetry
fftmap = fc.fft_map(symmetry_flags=symmetry_flags)
# rmup = fftmap.real_map_unpadded()
rm = fftmap.real_map().deep_copy()
maptbx.unpad_in_place(rm)
mmm = rm.as_1d().min_max_mean()
print(mmm.min, mmm.max, mmm.mean)
# rmup = fftmap.real_map_unpadded()
# print (dir(rm))
print("full size:", fftmap.real_map().accessor().focus())
print(rm[0, 0, 0])
# print (type(rm))
# print (dir(rm))
# STOP()
# print(rmup[0,0,0])
amap0 = asymmetric_map(xrs.space_group().type(), rm)
# print(dir(amap0))
mmm = amap0.data().as_1d().min_max_mean()
print(mmm.min, mmm.max, mmm.mean)
amap_data = amap0.data()
write_ccp4_map('amap.ccp4', xrs.unit_cell(), xrs.space_group(), amap_data)
write_ccp4_map('rm.ccp4', xrs.unit_cell(), xrs.space_group(), rm)
# for i in range(50):
# print(i, amap_data[i,0,0])
exp_map = amap0.symmetry_expanded_map()
print(exp_map[0, 0, 0])
# for i in range(32):
# for j in range(32):
# for k in range(32):
# assert approx_equal(rm[i,j,k], exp_map[i,j,k])
# print(dir(amap0))
# STOP()
# This produces 2 separate blobs
sg = xrs.space_group()
print(dir(sg))
print(sg.all_ops())
print(sg.info())
print("amap0 size:", amap0.data().accessor().focus())
# STOP()
print(type(amap0.data()))
threshold = 0.
preprocess_against_shallow = True
print('threshold:', threshold)
print('preprocess_against_shallow', preprocess_against_shallow)
t0 = time()
co_amap = maptbx.connectivity(
map_data=amap0.data(),
# threshold=threshold,
# space_group=xrs.space_group(),
# uc_dimensions=exp_map.accessor().focus(),
# wrapping=False,
preprocess_against_shallow=preprocess_against_shallow)
t1 = time()
print('amap time:', t1 - t0)
original_regions = list(co_amap.regions())
print('start regions:', original_regions)
print('max coords', list(co_amap.maximum_coors()))
print('max vals', list(co_amap.maximum_values()))
# print(dir(exp_map))
print(type(exp_map))
print("exp_map size:", exp_map.accessor().focus())
t0 = time()
co_full = maptbx.connectivity(
map_data=rm,
threshold=threshold,
wrapping=False,
preprocess_against_shallow=preprocess_against_shallow)
t1 = time()
print('full time:', t1 - t0)
original_regions = list(co_full.regions())
print('start regions:', original_regions)
print('max coords', list(co_full.maximum_coors()))
print('max vals', list(co_full.maximum_values()))
# STOP()
# co.experiment_with_symmetry(
# space_group=xrs.space_group(),
# uc_dims=exp_map.accessor().focus())
co_full.merge_symmetry_related_regions(space_group=xrs.space_group(),
uc_dims=exp_map.accessor().focus())
new_regions = list(co_full.regions())
print('new regions:', new_regions)
print('max coords', list(co_full.maximum_coors()))
print('max vals', list(co_full.maximum_values()))
def __init__(self,
xray_structure,
step,
volume_cutoff,
f_obs,
f_calc=None,
log=sys.stdout,
write_masks=False):
adopt_init_args(self, locals())
#
self.dsel = f_obs.d_spacings().data() >= 0
self.miller_array = f_obs.select(self.dsel)
#
self.crystal_symmetry = self.xray_structure.crystal_symmetry()
# compute mask in p1 (via ASU)
self.crystal_gridding = maptbx.crystal_gridding(
unit_cell=xray_structure.unit_cell(),
space_group_info=xray_structure.space_group_info(),
symmetry_flags=maptbx.use_space_group_symmetry,
step=step)
self.n_real = self.crystal_gridding.n_real()
# XXX Where do we want to deal with H and occ==0?
mask_p1 = mmtbx.masks.mask_from_xray_structure(
xray_structure=xray_structure,
p1=True,
for_structure_factors=True,
n_real=self.n_real,
in_asu=False).mask_data
maptbx.unpad_in_place(map=mask_p1)
self.solvent_content = 100. * mask_p1.count(1) / mask_p1.size()
if (write_masks):
write_map_file(crystal_symmetry=xray_structure.crystal_symmetry(),
map_data=mask_p1,
file_name="mask_whole.mrc")
# conn analysis
co = maptbx.connectivity(map_data=mask_p1,
threshold=0.01,
preprocess_against_shallow=True,
wrapping=True)
co.merge_symmetry_related_regions(
space_group=xray_structure.space_group())
del mask_p1
self.conn = co.result().as_double()
z = zip(co.regions(), range(0, co.regions().size()))
sorted_by_volume = sorted(z, key=lambda x: x[0], reverse=True)
f_mask_data_0 = flex.complex_double(f_obs.data().size(), 0)
FM = OrderedDict()
self.FV = OrderedDict()
self.mc = None
diff_map = None
mean_diff_map = None
self.regions = OrderedDict()
print(
" volume_p1 uc(%) volume_asu id mFo-DFc: min,max,mean,sd",
file=log)
# Check if self.anomaly
self.anomaly = False
if (len(sorted_by_volume) > 2):
uc_fractions = [
round(p[0] * 100. / self.conn.size(), 0)
for p in sorted_by_volume[1:]
]
if (uc_fractions[0] / 4 < uc_fractions[1]): self.anomaly = True
#
for i_seq, p in enumerate(sorted_by_volume):
v, i = p
# skip macromolecule
if (i == 0): continue
# skip small volume
volume = v * step**3
uc_fraction = v * 100. / self.conn.size()
if (volume_cutoff is not None):
if volume < volume_cutoff: continue
selection = self.conn == i
mask_i_asu = self.compute_i_mask_asu(selection=selection,
volume=volume)
volume_asu = (mask_i_asu > 0).count(True) * step**3
if (i_seq == 1 or uc_fraction > 5):
f_mask_i = self.compute_f_mask_i(mask_i_asu)
if (not self.anomaly):
f_mask_data_0 += f_mask_i.data()
if (uc_fraction < 5 and diff_map is None and not self.anomaly):
diff_map = self.compute_diff_map(f_mask_data=f_mask_data_0)
mi, ma, me, sd = None, None, None, None
if (diff_map is not None):
blob = diff_map.select(selection.iselection())
mean_diff_map = flex.mean(
diff_map.select(selection.iselection()))
mi, ma, me = flex.min(blob), flex.max(blob), flex.mean(blob)
sd = blob.sample_standard_deviation()
print("%12.3f" % volume,
"%8.4f" % round(uc_fraction, 4),
"%12.3f" % volume_asu,
"%3d" % i,
"%7s" % str(None) if diff_map is None else
"%7.3f %7.3f %7.3f %7.3f" % (mi, ma, me, sd),
file=log)
if (uc_fraction < 1 and mean_diff_map is not None
and mean_diff_map <= 0):
continue
self.regions[i_seq] = group_args(id=i,
i_seq=i_seq,
volume=volume,
uc_fraction=uc_fraction,
diff_map=group_args(mi=mi,
ma=ma,
me=me,
sd=sd))
if (not (i_seq == 1 or uc_fraction > 5)):
f_mask_i = self.compute_f_mask_i(mask_i_asu)
FM.setdefault(round(volume, 3), []).append(f_mask_i.data())
self.FV[f_mask_i] = [round(volume, 3), round(uc_fraction, 1)]
#
f_mask_0 = f_obs.customized_copy(data=f_mask_data_0)
#
self.f_mask_0 = None
if (not self.anomaly):
self.f_mask_0 = f_obs.customized_copy(data=f_mask_data_0)
self.do_mosaic = False
if (len(self.FV.keys()) > 1):
self.do_mosaic = True
def __init__(self,
xray_structure,
step,
volume_cutoff=None,
mean_diff_map_threshold=None,
compute_whole=False,
largest_only=False,
wrapping=True,
f_obs=None,
r_sol=1.1,
r_shrink=0.9,
f_calc=None,
log=None,
write_masks=False):
adopt_init_args(self, locals())
#
self.d_spacings = f_obs.d_spacings().data()
self.sel_3inf = self.d_spacings >= 3
self.miller_array = f_obs.select(self.sel_3inf)
#
self.crystal_symmetry = self.xray_structure.crystal_symmetry()
# compute mask in p1 (via ASU)
self.crystal_gridding = maptbx.crystal_gridding(
unit_cell=xray_structure.unit_cell(),
space_group_info=xray_structure.space_group_info(),
symmetry_flags=maptbx.use_space_group_symmetry,
step=step)
self.n_real = self.crystal_gridding.n_real()
# XXX Where do we want to deal with H and occ==0?
mask_p1 = mmtbx.masks.mask_from_xray_structure(
xray_structure=xray_structure,
p1=True,
for_structure_factors=True,
solvent_radius=r_sol,
shrink_truncation_radius=r_shrink,
n_real=self.n_real,
in_asu=False).mask_data
maptbx.unpad_in_place(map=mask_p1)
self.f_mask_whole = None
if (compute_whole):
mask = asu_map_ext.asymmetric_map(
xray_structure.crystal_symmetry().space_group().type(),
mask_p1).data()
self.f_mask_whole = self._inflate(
self.miller_array.structure_factors_from_asu_map(
asu_map_data=mask, n_real=self.n_real))
self.solvent_content = 100. * mask_p1.count(1) / mask_p1.size()
if (write_masks):
write_map_file(crystal_symmetry=xray_structure.crystal_symmetry(),
map_data=mask_p1,
file_name="mask_whole.mrc")
# conn analysis
co = maptbx.connectivity(map_data=mask_p1,
threshold=0.01,
preprocess_against_shallow=False,
wrapping=wrapping)
co.merge_symmetry_related_regions(
space_group=xray_structure.space_group())
del mask_p1
self.conn = co.result().as_double()
z = zip(co.regions(), range(0, co.regions().size()))
sorted_by_volume = sorted(z, key=lambda x: x[0], reverse=True)
#
f_mask_data_0 = flex.complex_double(f_obs.data().size(), 0)
f_mask_data = flex.complex_double(f_obs.data().size(), 0)
self.FV = OrderedDict()
self.mc = None
diff_map = None
mean_diff_map = None
self.regions = OrderedDict()
self.f_mask_0 = None
self.f_mask = None
#
if (log is not None):
print(" # volume_p1 uc(%) mFo-DFc: min,max,mean,sd",
file=log)
#
for i_seq, p in enumerate(sorted_by_volume):
v, i = p
# skip macromolecule
if (i == 0): continue
# skip small volume
volume = v * step**3
uc_fraction = v * 100. / self.conn.size()
if (volume_cutoff is not None):
if volume < volume_cutoff: continue
self.regions[i_seq] = group_args(id=i,
i_seq=i_seq,
volume=volume,
uc_fraction=uc_fraction)
selection = self.conn == i
mask_i_asu = self.compute_i_mask_asu(selection=selection,
volume=volume)
volume_asu = (mask_i_asu > 0).count(True) * step**3
if (uc_fraction >= 1):
f_mask_i = self.compute_f_mask_i(mask_i_asu)
f_mask_data_0 += f_mask_i.data()
elif (largest_only):
break
if (uc_fraction < 1 and diff_map is None):
diff_map = self.compute_diff_map(f_mask_data=f_mask_data_0)
mi, ma, me, sd = None, None, None, None
if (diff_map is not None):
blob = diff_map.select(selection.iselection())
mean_diff_map = flex.mean(
diff_map.select(selection.iselection()))
mi, ma, me = flex.min(blob), flex.max(blob), flex.mean(blob)
sd = blob.sample_standard_deviation()
if (log is not None):
print("%3d" % i_seq,
"%12.3f" % volume,
"%8.4f" % round(uc_fraction, 4),
"%7s" % str(None) if diff_map is None else
"%7.3f %7.3f %7.3f %7.3f" % (mi, ma, me, sd),
file=log)
if (mean_diff_map_threshold is not None
and mean_diff_map is not None
and mean_diff_map <= mean_diff_map_threshold):
continue
f_mask_i = self.compute_f_mask_i(mask_i_asu)
f_mask_data += f_mask_i.data()
self.FV[f_mask_i] = [round(volume, 3), round(uc_fraction, 1)]
#
self.f_mask_0 = f_obs.customized_copy(data=f_mask_data_0)
self.f_mask = f_obs.customized_copy(data=f_mask_data)
self.do_mosaic = False
# Determine number of secondary regions
self.n_regions = len(self.FV.values())
if (self.n_regions > 1):
self.do_mosaic = True
def exercise_symmetry_related_regions():
pdb_str = """
CRYST1 10.000 10.000 10.000 90.00 90.00 90.00 P 4
HETATM 1 C C 1 2.000 2.000 2.000 1.00 20.00 C
HETATM 2 C C 2 4.000 4.000 4.000 1.00 20.00 C
END
"""
pdb_inp = iotbx.pdb.input(source_info=None, lines=pdb_str)
xrs = pdb_inp.xray_structure_simple()
# xrs.show_summary()
d_min = 1.
fc = xrs.structure_factors(d_min=d_min).f_calc()
symmetry_flags = maptbx.use_space_group_symmetry
fftmap = fc.fft_map(symmetry_flags=symmetry_flags)
rmup = fftmap.real_map_unpadded()
# print ('rmup size', rmup.accessor().focus())
# This produces 4 separate blobs
co = maptbx.connectivity(map_data=rmup,
threshold=400,
wrapping=False,
preprocess_against_shallow=False)
original_regions = list(co.regions())
# print ('regions', original_regions)
assert len(original_regions) == 5
beg_mask = co.result()
# dv_mask = co.volume_cutoff_mask(0).as_double() ???
# write_ccp4_map('volume_mask_1000.ccp4', fc.unit_cell(), fc.space_group(), dv_mask)
co.merge_symmetry_related_regions(space_group=xrs.space_group())
new_mask = co.result()
assert beg_mask.count(0) == new_mask.count(0)
assert beg_mask.count(1) + beg_mask.count(3) == new_mask.count(1)
assert beg_mask.count(2) + beg_mask.count(4) == new_mask.count(2)
assert sum(original_regions[1:]) == sum(original_regions[1:])
new_regions = list(co.regions())
assert len(new_regions) == 3
assert list(co.maximum_values()) == []
assert list(co.maximum_coors()) == []
# ======================================================================
# At this threshold 2 carbons merge. But one of the blob is cutted,
# therefore producing 3 separate regions in unit cell
co = maptbx.connectivity(
map_data=rmup,
# threshold=1000,
threshold=1.1,
wrapping=False,
preprocess_against_shallow=True)
original_regions = list(co.regions())
assert len(original_regions) == 4
# print ('regions', original_regions)
beg_mask = co.result()
# Particular numbers here seem to be platform-dependent
# These should work on Mac
# assert beg_mask.count(0) == 29019
# assert beg_mask.count(1) == 1885
# assert beg_mask.count(2) == 1714
# assert beg_mask.count(3) == 150
# assert original_regions == [29019, 1885, 1714, 150]
co.merge_symmetry_related_regions(space_group=xrs.space_group())
new_mask = co.result()
# assert new_mask.count(0) == 29019
# assert new_mask.count(1) == 3749
assert beg_mask.count(0) == new_mask.count(0)
assert beg_mask.count(1) + beg_mask.count(2) + beg_mask.count(
3) == new_mask.count(1)
new_regions = list(co.regions())
assert len(new_regions) == 2
# print('new regs', new_regions)
# assert new_regions == [29019, 3749]
assert list(co.maximum_values()) == []
assert list(co.maximum_coors()) == []
def exercise_preprocess_against_shallow():
# case 1: simple
cmap = flex.double(flex.grid(30, 30, 30))
cmap.fill(1)
for i in range(10, 20):
for j in range(10, 20):
for k in range(10, 20):
cmap[i, j, k] = 10
for i in range(10, 20):
cmap[i, 5, 5] = 10
co = maptbx.connectivity(map_data=cmap, threshold=5)
minb, maxb = co.get_blobs_boundaries_tuples()
assert minb == [(0, 0, 0), (10, 5, 5), (10, 10, 10)]
assert maxb == [(29, 29, 29), (19, 5, 5), (19, 19, 19)]
co = maptbx.connectivity(map_data=cmap,
threshold=5,
preprocess_against_shallow=True)
minb, maxb = co.get_blobs_boundaries_tuples()
assert minb == [(0, 0, 0), (10, 10, 10)]
assert maxb == [(29, 29, 29),
(19, 19, 19)] # note dissapearance of (10,5,5)(19,5,5)
# check new map values
for i in range(10, 20):
assert approx_equal(cmap[i, 5, 5], 4)
# case 2: wrapping
cmap = flex.double(flex.grid(30, 30, 30))
cmap.fill(1)
for i in range(10, 20):
for j in range(10, 20):
for k in range(10, 20):
cmap[i, j, k] = 10
for i in range(10, 20):
for j in range(10, 20):
cmap[i, j, 0] = 10
cmap[i, j, 29] = 10
# standard, no wrap, 4 regions
co = maptbx.connectivity(map_data=cmap,
threshold=5,
wrapping=False,
preprocess_against_shallow=False)
minb, maxb = co.get_blobs_boundaries_tuples()
assert minb == [(0, 0, 0), (10, 10, 0), (10, 10, 10), (10, 10, 29)]
assert maxb == [(29, 29, 29), (19, 19, 0), (19, 19, 19), (19, 19, 29)]
# 2 small regions merged
co = maptbx.connectivity(map_data=cmap,
threshold=5,
wrapping=True,
preprocess_against_shallow=False)
minb, maxb = co.get_blobs_boundaries_tuples()
assert minb == [(0, 0, 0), (10, 10, 0), (10, 10, 10)]
assert maxb == [(29, 29, 29), (19, 19, 29), (19, 19, 19)]
# with wrapping the region preserved
co = maptbx.connectivity(map_data=cmap,
threshold=5,
wrapping=True,
preprocess_against_shallow=True)
minb, maxb = co.get_blobs_boundaries_tuples()
assert minb == [(0, 0, 0), (10, 10, 0), (10, 10, 10)]
assert maxb == [(29, 29, 29), (19, 19, 29), (19, 19, 19)]
# without wrapping - no
co = maptbx.connectivity(map_data=cmap,
threshold=5,
wrapping=False,
preprocess_against_shallow=True)
minb, maxb = co.get_blobs_boundaries_tuples()
assert minb == [(0, 0, 0), (10, 10, 10)]
assert maxb == [(29, 29, 29), (19, 19, 19)]
# case 3: blob has a spike that needs to be 'shaved off'
cmap = flex.double(flex.grid(30, 30, 30))
cmap.fill(1)
for i in range(10, 20):
for j in range(10, 20):
for k in range(10, 20):
cmap[i, j, k] = 10
for i in range(0, 10):
cmap[i, 15, 15] = 10
co = maptbx.connectivity(map_data=cmap,
threshold=5,
preprocess_against_shallow=False)
volumes = list(co.regions())
assert volumes == [25990, 1010]
co = maptbx.connectivity(map_data=cmap,
threshold=5,
preprocess_against_shallow=True)
volumes = list(co.regions())
assert volumes == [26000, 1000]
for i in range(0, 10):
assert approx_equal(cmap[i, 15, 15], 4)
# case 4: need at least two passes
cmap = flex.double(flex.grid(30, 30, 30))
cmap.fill(1)
cmap[10, 10, 10] = 10
cmap[9, 10, 10] = 10
cmap[11, 10, 10] = 10
cmap[10, 9, 10] = 10
cmap[10, 11, 10] = 10
cmap[10, 10, 9] = 10
cmap[10, 10, 11] = 10
co = maptbx.connectivity(map_data=cmap,
threshold=5,
preprocess_against_shallow=False)
volumes = list(co.regions())
assert volumes == [26993, 7]
co = maptbx.connectivity(map_data=cmap,
threshold=5,
preprocess_against_shallow=True)
volumes = list(co.regions())
assert volumes == [27000]
assert co.preprocessing_changed_voxels == 7
assert co.preprocessing_n_passes == 3
def run_group(symbol, preprocess_against_shallow):
group = space_group_info(symbol)
print("\n== space group %d"%symbol)
xrs = random_structure.xray_structure(
space_group_info = group,
volume_per_atom = 15.,
general_positions_only = False,
elements = ('C', 'N', 'O', 'H')*31,
min_distance = 1.0)
sgt = xrs.space_group().type()
#
cg = maptbx.crystal_gridding(
unit_cell = xrs.unit_cell(),
space_group_info = xrs.space_group_info(),
symmetry_flags = maptbx.use_space_group_symmetry,
step = 0.4)
n_real = cg.n_real()
mask_p1 = mmtbx.masks.mask_from_xray_structure(
xray_structure = xrs,
p1 = True,
for_structure_factors = True,
n_real = n_real,
in_asu = False).mask_data
maptbx.unpad_in_place(map=mask_p1)
assert flex.min(mask_p1)==0
assert flex.max(mask_p1)==1
#
co = maptbx.connectivity(
map_data = mask_p1,
threshold = 0.01,
preprocess_against_shallow = preprocess_against_shallow,
wrapping = True)
#
print("Regions in P1")
regions_p1 = list(co.regions())
s1 = flex.sum(flex.int(regions_p1))
print(regions_p1, s1)
conn_map_p1 = co.result().as_double()
print(flex.min(conn_map_p1), flex.max(conn_map_p1))
#
print("Merge symmetry related")
co.merge_symmetry_related_regions(space_group = xrs.space_group())
conn_map_p1_merged = co.result().as_double()
regions_p1_merged = list(co.regions())
s2 = flex.sum(flex.int(regions_p1_merged))
print(list(regions_p1_merged), s2)
amap = asu_map_ext.asymmetric_map(sgt, conn_map_p1_merged)
conn_map_asu = amap.data()
conn_map_p1_restored = amap.symmetry_expanded_map()
print(flex.min(conn_map_asu), flex.max(conn_map_asu))
#
mask_p1_1 = conn_map_p1_restored.set_selected(conn_map_p1_restored>0.01, 1)
maptbx.unpad_in_place(map=mask_p1_1)
co = maptbx.connectivity(
map_data = mask_p1_1,
threshold = 0.01,
preprocess_against_shallow = preprocess_against_shallow,
wrapping = True)
print("Restored")
regions_p1_restored = list(co.regions())
s3 = flex.sum(flex.int(regions_p1_restored))
print(regions_p1_restored, s3)
conn_map_p1_restored = co.result().as_double()
print(flex.min(conn_map_p1_restored), flex.max(conn_map_p1_restored))
assert regions_p1 == regions_p1_restored
#
assert s1 == s2
assert s2 == s3
def __init__(self,
miller_array,
xray_structure,
step,
volume_cutoff,
f_obs=None,
r_free_flags=None,
f_calc=None,
write_masks=False):
adopt_init_args(self, locals())
assert [f_obs, f_calc, r_free_flags].count(None) in [0, 3]
self.crystal_symmetry = self.xray_structure.crystal_symmetry()
# compute mask in p1 (via ASU)
self.crystal_gridding = maptbx.crystal_gridding(
unit_cell=xray_structure.unit_cell(),
space_group_info=xray_structure.space_group_info(),
symmetry_flags=maptbx.use_space_group_symmetry,
step=step)
self.n_real = self.crystal_gridding.n_real()
mask_p1 = mmtbx.masks.mask_from_xray_structure(
xray_structure=xray_structure,
p1=True,
for_structure_factors=True,
n_real=self.n_real,
in_asu=False).mask_data
maptbx.unpad_in_place(map=mask_p1)
solvent_content = 100. * mask_p1.count(1) / mask_p1.size()
if (write_masks):
write_map_file(crystal_symmetry=xray_structure.crystal_symmetry(),
map_data=mask_p1,
file_name="mask_whole.mrc")
# conn analysis
co = maptbx.connectivity(map_data=mask_p1,
threshold=0.01,
preprocess_against_shallow=True,
wrapping=True)
del mask_p1
self.conn = co.result().as_double()
z = zip(co.regions(), range(0, co.regions().size()))
sorted_by_volume = sorted(z, key=lambda x: x[0], reverse=True)
f_mask_data = flex.complex_double(miller_array.data().size(), 0)
f_mask_data_0 = flex.complex_double(miller_array.data().size(), 0)
#f_masks = []
FM = OrderedDict()
diff_map = None
mean_diff_map = None
print(" volume_p1 uc(%) volume_asu id <mFo-DFc>")
for p in sorted_by_volume:
v, i = p
volume = v * step**3
uc_fraction = v * 100. / self.conn.size()
if (volume_cutoff is not None):
if volume < volume_cutoff: continue
if (i == 0): continue
selection = self.conn == i
mask_i_asu = self.compute_i_mask_asu(selection=selection,
volume=volume)
volume_asu = (mask_i_asu > 0).count(True) * step**3
if (volume_asu < 1.e-6): continue
if (i == 1 or uc_fraction > 5):
f_mask_i = miller_array.structure_factors_from_asu_map(
asu_map_data=mask_i_asu, n_real=self.n_real)
f_mask_data_0 += f_mask_i.data()
f_mask_data += f_mask_i.data()
if (uc_fraction < 5 and diff_map is None):
diff_map = self.compute_diff_map(f_mask_data=f_mask_data_0)
if (diff_map is not None):
mean_diff_map = flex.mean(
diff_map.select(selection.iselection()))
print(
"%12.3f" % volume, "%8.4f" % round(uc_fraction, 4),
"%12.3f" % volume_asu, "%3d" % i, "%7s" %
str(None) if diff_map is None else "%7.3f" % mean_diff_map)
#if(mean_diff_map is not None and mean_diff_map<=0): continue
if (not (i == 1 or uc_fraction > 5)):
f_mask_i = miller_array.structure_factors_from_asu_map(
asu_map_data=mask_i_asu, n_real=self.n_real)
f_mask_data += f_mask_i.data()
FM.setdefault(round(volume, 3), []).append(f_mask_i.data())
# group asu pices corresponding to the same region in P1
F_MASKS = []
for k, v in zip(FM.keys(), FM.values()):
tmp = flex.complex_double(miller_array.data().size(), 0)
for v_ in v:
tmp += v_
F_MASKS.append(miller_array.customized_copy(data=tmp))
#
f_mask = miller_array.customized_copy(data=f_mask_data)
#
self.f_mask = f_mask
self.f_masks = F_MASKS