def write_map(self, filename='converted.map'):
if(self.map is None):
hex=False
self.zga = math.zernike_grid( self.np_on_grid, self.nmax, hex )
self.zga.load_coefs( self.nlm_array.nlm(), self.nlm_array.coefs() )
self.map = flex.abs( self.zga.f() )
xplor_map_type( self.map, self.np_on_grid, self.rmax, file_name=filename )
def __init__(self,
data,
xplor_file,
rmax,
qmax=0.15,
nmax=20,
np_on_grid=30,
prefix='prefix',
splat_range=1,
n_trial=5,
fraction=0.9,
pdb_nlm=None,
nbr_dist=2):
self.raw_data = data
self.rmax = rmax / fraction
self.fraction = fraction
self.qmax = qmax
self.nmax = nmax
self.nbr_dist = nbr_dist
self.pdb_nlm = pdb_nlm
self.np_on_grid = np_on_grid
self.zga = math.zernike_grid(self.np_on_grid, self.nmax, False)
self.n = self.np_on_grid * 2 + 1
self.n2 = self.n**2
self.n3 = self.n2 * self.n
self.prefix = prefix + '_'
self.splat_range = splat_range
self.start_model = None
neighbors = [(-1, 0, 0), (1, 0, 0), (0, -1, 0), (0, 1, 0), (0, 0, -1),
(0, 0, 1)]
self.neighbors = flex.int()
for n in neighbors:
self.neighbors.append(self.convert_indx_3_1(n))
self.build_nbr_list(self.nbr_dist)
print "neighbor distance: ", self.nbr_dist
self.build_sphere_list()
#### Labels for different regions ####
self.solvent_label = 0
self.molecule_label = 1
self.surface_label = 2
self.nlm_array = math.nlm_array(nmax)
self.nlm = self.nlm_array.nlm()
self.counter = 0
self.n_accept = 0
self.bandwidth = min(smath.pi / rmax / 2.0, 0.01)
# self.data = reduce_raw_data(self.raw_data,self.qmax, self.bandwidth,level=0.00001 )
self.data = self.raw_data
self.data.i = self.data.i / self.data.i[0]
self.data.s = self.data.i
self.scale_2_expt = self.data.i[0]
self.build_starting_model()
self.mark_mod_core_region0(splat_range)
#self.mark_mod_core_region(splat_range)
self.n_trial = n_trial
self.finals = []
for ii in range(n_trial):
self.refine(ii)
self.finals.append(self.best_nlm_coefs.deep_copy())
self.pair_align()
def tst_zernike_grid(skip_iteration_probability=0.95):
#THIS TEST TAKES A BIT OF TIME
M=20
N=4
ddd = (M*2+1)
zga = math.zernike_grid(M,N,False)
zgb = math.zernike_grid(M,N,False)
xyz = zga.xyz()
coefs = zga.coefs()
nlm = zga.nlm()
import random
rng = random.Random(x=None)
for ii in range(nlm.size()):
for jj in range(ii+1,nlm.size()):
if (rng.random() < skip_iteration_probability):
continue
coefsa = coefs*0.0
coefsb = coefs*0.0
coefsa[ii]=1.0+1.0j
coefsb[jj]=1.0+1.0j
zga.load_coefs( nlm, coefsa )
zgb.load_coefs( nlm, coefsb )
fa = zga.f()
fb = zgb.f()
prodsum = flex.sum(fa*flex.conj(fb) )/xyz.size()
prodsuma= flex.sum(fa*flex.conj(fa) )/xyz.size()
prodsumb= flex.sum(fb*flex.conj(fb) )/xyz.size()
t1 = abs(prodsum)
t2 = abs(prodsuma)
t3 = abs(prodsumb)
t1 = 100.0*(t1/t2)
t2 = 100.0*(abs(t2-t3)/t3)
# unfortunately, this numerical integration scheme is not optimal. For certain
# combinations of nlm, we see significant non-orthogonality that reduces when
# we increase the number of points. A similar behavior is seen in the radial
# part of the Zernike polynome. If we compile withiout the radial function, similar
# behavior is seen using only the spherical harmonics functions.
# For this reason, the liberal limts set below are ok
assert t1<2.0
assert t2<5.0
def tst_zernike_grid(skip_iteration_probability=0.95):
#THIS TEST TAKES A BIT OF TIME
M=20
N=4
ddd = (M*2+1)
zga = math.zernike_grid(M,N,False)
zgb = math.zernike_grid(M,N,False)
xyz = zga.xyz()
coefs = zga.coefs()
nlm = zga.nlm()
import random
rng = random.Random(x=None)
for ii in range(nlm.size()):
for jj in range(ii+1,nlm.size()):
if (rng.random() < skip_iteration_probability):
continue
coefsa = coefs*0.0
coefsb = coefs*0.0
coefsa[ii]=1.0+1.0j
coefsb[jj]=1.0+1.0j
zga.load_coefs( nlm, coefsa )
zgb.load_coefs( nlm, coefsb )
fa = zga.f()
fb = zgb.f()
prodsum = flex.sum(fa*flex.conj(fb) )/xyz.size()
prodsuma= flex.sum(fa*flex.conj(fa) )/xyz.size()
prodsumb= flex.sum(fb*flex.conj(fb) )/xyz.size()
t1 = abs(prodsum)
t2 = abs(prodsuma)
t3 = abs(prodsumb)
t1 = 100.0*(t1/t2)
t2 = 100.0*(abs(t2-t3)/t3)
# unfortunately, this numerical integration scheme is not optimal. For certain
# combinations of nlm, we see significant non-orthogonality that reduces when
# we increase the number of points. A similar behavior is seen in the radial
# part of the Zernike polynome. If we compile withiout the radial function, similar
# behavior is seen using only the spherical harmonics functions.
# For this reason, the liberal limts set below are ok
assert t1<2.0
assert t2<5.0
def write_bead( self, filename='converted.pdb'):
self.n_bds = int(self.rmax/(3.6))+1
hex = True
self.zga_hex = math.zernike_grid( self.n_bds, self.nmax, hex)
self.zga_hex.load_coefs( self.nlm_array.nlm(), self.nlm_array.coefs() )
self.map_hex = flex.abs( self.zga_hex.f() )
max_value = flex.max( self.map_hex )
threshold = max_value*0.3
select = flex.bool( self.map_hex.as_1d() >= threshold )
xyz = self.zga_hex.xyz().select( select )
xyz = (xyz+(1.0,1.0,1.0)) * self.rmax
res_id = 1
out = open( filename, 'w')
for x,y,z in xyz:
print>>out, "ATOM %5d CA ASP %4d %8.3f%8.3f%8.3f 1.00 1.00"%(res_id, res_id, x, y, z)
res_id = res_id + 1
out.close()
def build_map(nmax,
rmax,
coefs,
codes,
model_indx,
pdb_models,
prefix,
clusters=None,
fract=0.9,
type='.ccp4'):
global stdfile
global outfilelog
rmax_over_fraction = rmax / fract
np_on_grid = 30
zga = math.zernike_grid(np_on_grid, nmax, False)
ref_nlm_array = math.nlm_array(nmax)
mov_nlm_array = math.nlm_array(nmax)
nlm = ref_nlm_array.nlm()
nlm_total = ref_nlm_array.coefs().size()
top_cc = flex.double()
top_ids = []
ntop = model_indx.size()
rank = 0
ave_c = flex.complex_double(nlm_total, 0)
aligned_coefs = []
map_files = []
levels = []
scale_model = False
if ((pdb_models is not None) and (pdb_models[0].rmax > rmax)):
scale_model = True
external_rmax = pdb_models[0].rmax
grid = build_3d_grid(np_on_grid, rmax_over_fraction)
with open(stdfile, "a") as log:
print >> log, "Rank PDB_code cc (to the given model or the first model):"
print "Rank PDB_code cc (to the given model or the first model):"
if (pdb_models is not None):
ref_nlm_array.load_coefs(nlm, pdb_models[0].nlm_coef)
fraction = 0
else:
c = coefs[model_indx[0]][0:nlm_total]
ave_c = c.deep_copy()
ref_nlm_array.load_coefs(nlm, c)
rank = 1
if prefix != None:
filename = prefix + "m" + str(rank) + "_" + codes[model_indx[0]]
#filename = os.path.join(prefix,"m"+str(rank)+"_"+codes[ model_indx[0] ])
else:
filename = "m" + str(rank) + "_" + codes[model_indx[0]]
map_files.append(filename + type)
fraction, map = write_map(zga,
nlm,
c,
np_on_grid,
rmax_over_fraction,
filename,
type=type)
#level = flex.max(map)/3.0
print "map in search pdb.py: \n"
print map
level = map.standard_deviation_of_the_sample()
print "level in search pdb: ", level
levels.append(level)
top_cc.append(1.0) # refering to itself
if (scale_model):
c = get_moments_for_scaled_model(map, np_on_grid, grid, nmax, rmax,
external_rmax)
with open(stdfile, "a") as log:
print >> log, rank, codes[model_indx[0]]
print rank, codes[model_indx[0]]
aligned_coefs.append(c.deep_copy()) # save the aligned nlm coefs
mean_frac = fraction
mean_sqr_frac = fraction * fraction
for ii in model_indx[rank:]:
rank = rank + 1
c = coefs[ii][0:nlm_total]
mov_nlm_array.load_coefs(nlm, c)
align_obj = fft_align.align(ref_nlm_array,
mov_nlm_array,
nmax=nmax,
refine=True)
new_c = align_obj.moving_nlm.coefs()
if prefix != None:
#filename = os.path.join(prefix,"m"+str(rank)+"_"+codes[ii])
filename = prefix + "m" + str(rank) + "_" + codes[ii]
print "**********************************"
print "outfilelog: ", outfilelog
with open(outfilelog, "a") as f:
print >> f, filename + ".ccp4"
else:
filename = "m" + str(rank) + "_" + codes[ii]
with open(outfilelog, "a") as f:
print >> f, filename + ".ccp4"
map_files.append(filename + type)
fraction, map = write_map(zga,
nlm,
new_c,
np_on_grid,
rmax_over_fraction,
filename,
type=type)
if (scale_model):
c = get_moments_for_scaled_model(map, np_on_grid, grid, nmax, rmax,
external_rmax)
mov_nlm_array.load_coefs(nlm, c)
align_obj = fft_align.align(ref_nlm_array,
mov_nlm_array,
nmax=nmax,
refine=True)
new_c = align_obj.moving_nlm.coefs()
fraction, map = write_map(zga,
nlm,
new_c,
np_on_grid,
rmax_over_fraction,
filename,
type=type)
level = flex.max(map) / 3.0
levels.append(level)
cc = align_obj.get_cc()
with open(stdfile, "a") as log:
print >> log, "%2d %5s %5.3f" % (rank, codes[ii], cc)
print "%2d %5s %5.3f" % (rank, codes[ii], cc)
top_cc.append(cc)
top_ids.append(codes[ii])
ave_c = ave_c + new_c
aligned_coefs.append(new_c.deep_copy()) # save the aligned nlm coefs
mean_frac = mean_frac + fraction
mean_sqr_frac = mean_sqr_frac + fraction * fraction
sphere_volume = rmax_over_fraction**3.0 * 8.0 # cube with d=2.0*r
mean_frac = mean_frac / ntop
sigma_frac = smath.sqrt(mean_sqr_frac / ntop - mean_frac * mean_frac)
with open(stdfile, "a") as log:
print >> log, "Volume is ", mean_frac * sphere_volume, "+/-", sigma_frac * sphere_volume, "(A^3)"
print "Volume is ", mean_frac * sphere_volume, "+/-", sigma_frac * sphere_volume, "(A^3)"
#### Write average map ####
ave_maps = []
ave_levels = []
ave_cc = []
cluster_ids = [1] * len(model_indx) # initialize cluster_ids
# filename = "ave_map"
# map_files.append( filename+type )
# fraction, map=write_map( zga, nlm, ave_c/ntop, np_on_grid, rmax_over_fraction, filename, type=type )
# levels.append( flex.max(map)/3.0 )
# if( len(clusters.nodes) == 1): return top_cc, top_ids, map_files, levels, [1]*len(map_files)
cluster_id = 1
with open(stdfile, "a") as log:
print >> log, "cc. between Cluster average and PDB model"
print "cc. between Cluster average and PDB model"
for node in clusters.nodes:
ave_c = ave_c * 0
coefs_list = []
for ii in node.leaf_eles:
ave_c = ave_c + aligned_coefs[ii]
cluster_ids[ii] = cluster_id
coefs_list.append(aligned_coefs[ii])
ave_c = ave_c / len(node.leaf_eles)
level_n = model_consistency.good2n(nmax,
coefs_list,
ave_c,
threshold=0.90,
outfile=stdfile)
with open(stdfile, "a") as log:
print >> log, "consistency level to order n: %d" % level_n
print "consistency level to order n: %d" % level_n
mov_nlm_array.load_coefs(nlm, ave_c)
align_obj = fft_align.align(ref_nlm_array,
mov_nlm_array,
nmax=nmax,
refine=True)
cc = align_obj.get_cc()
ave_cc.append(cc)
with open(stdfile, "a") as log:
print >> log, "cluster # ", cluster_id, "cc=", cc
print "cluster # ", cluster_id, "cc=", cc
if prefix == None:
filename = "ave_" + str(cluster_id)
with open(outfilelog, "a") as f:
print >> f, filename
else:
filename = prefix + "ave_" + str(cluster_id)
with open(outfilelog, "a") as f:
print >> f, filename + ".ccp4"
ave_maps.append(filename + type)
fraction, map = write_map(zga,
nlm,
ave_c,
np_on_grid,
rmax_over_fraction,
filename,
type=type)
ave_levels.append(flex.max(map) / 3.0)
with open(stdfile, "a") as log:
print >> log, "Averaged Model #%d Volume is %f (A^3)" % (
cluster_id, fraction * sphere_volume)
print "Averaged Model #%d Volume is %f (A^3)" % (cluster_id, fraction *
sphere_volume)
cluster_id = cluster_id + 1
# with open(stdfile,"a") as log:
# log.write("__END__")
return top_cc, top_ids, map_files, levels, cluster_ids, ave_maps, ave_levels, ave_cc
def zernike_moments(pdbfile,
nmax=20,
np=50,
fix_dx=False,
np_on_grid=20,
shift=False,
buildmap=False,
coef_out=True,
calc_intensity=True,
external_rmax=-1):
base = pdbfile.split('.')[0]
splat_range = 0
fraction = 0.9
default_dx = 0.7
uniform = True
pdbi = pdb.hierarchy.input(file_name=pdbfile)
# pdbi = iotbx.pdb.hierarchy.input(pdb_string='''ATOM 1 N ASP A 37 10.710 14.456 9.568 1.00 15.78 N''')
if (len(pdbi.hierarchy.models()) == 0):
return None, None, None
atoms = pdbi.hierarchy.models()[0].atoms()
# predefine some arrays we will need
atom_types = flex.std_string()
radius = flex.double()
b_values = flex.double()
occs = flex.double()
# xyz = flex.vec3_double()
xyz = get_xyz_using_split(pdbfile)
# keep track of the atom types we have encountered
# for atom in atoms:
# print"---==================atom=================---",atom.xyz
# if(not atom.hetero):
# xyz.append( atom.xyz )
# b_values.append( atom.b )
# occs.append( atom.occ )
#print time.time()
if (xyz.size() == 0):
return None, None, None
density = flex.double(xyz.size(), 1.0)
voxel_obj = math.sphere_voxel(np, splat_range, uniform, fix_dx,
external_rmax, default_dx, fraction, xyz,
density)
np = voxel_obj.np()
# print "*********************voxel_obj.rmax(): ",voxel_obj.rmax()
rmax = voxel_obj.rmax() / fraction
# print "*********************fraction: ",fraction
# print "*********************rmax=voxel_obj.rmax()/fraction: ",rmax
#print base, "RMAX: ", voxel_obj.rmax()
#print time.time()
grid_obj = math.sphere_grid(np, nmax)
pdb_out = False
grid_obj.clean_space(voxel_obj, pdb_out)
#print time.time(), "END GRID CONST"
grid_obj.construct_space_sum()
#print time.time(), "SS CALC"
mom_obj = math.zernike_moments(grid_obj, nmax)
#print time.time(), "END MOM CALC"
moments = mom_obj.moments()
if (coef_out):
nn = mom_obj.fnn()
easy_pickle.dump(base + '.nlm.pickle', moments.coefs())
easy_pickle.dump(base + '.nn.pickle', nn.coefs())
#
####### The following will be optional ###
if (shift):
shift = [rmax, rmax, rmax]
centered_xyz = voxel_obj.xyz() + shift
out_pdb_name = base + '_centered.pdb'
for a, xyz in zip(atoms, centered_xyz):
a.set_xyz(new_xyz=xyz)
pdbi.hierarchy.write_pdb_file(file_name=out_pdb_name,
open_append=False)
original_map = voxel_obj.map()
xplor_map_type(original_map, np, rmax, file_name=base + '_pdb.xplor')
######## Calculate Intnesity Profile ############
if (calc_intensity):
q_array = flex.double(range(51)) / 100.0
z_model = zm.zernike_model(moments, q_array, rmax, nmax)
nn = mom_obj.fnn()
intensity = z_model.calc_intensity(nn)
#iq_file = open(base+"_"+str(nmax)+"_"+str(int(rmax*fraction))+".zi", 'w')
iq_file = open(base + "_" + str(nmax) + ".zi", 'w')
for qq, ii in zip(q_array, intensity):
print >> iq_file, qq, ii
iq_file.close()
####### END of Intensity Calculation ###########
if (buildmap):
print("grid setting up...")
zga = math.zernike_grid(np_on_grid, nmax, False)
print("grid setting up...Done")
zga.load_coefs(moments.nlm(), moments.coefs())
print("start reconstruction")
map = flex.abs(zga.f())
print("finished reconstruction")
xplor_map_type(map, np_on_grid, rmax, file_name=base + '.xplor')
ccp4_map_type(map, np_on_grid, rmax, file_name=base + '.ccp4')
return mom_obj, voxel_obj, pdbi
def zernike_moments(mol2_file, nmax=20, np=50, uniform=True, fix_dx=False, np_on_grid=20, shift=False, buildmap=False, coef_out=True, calc_intensity=True, external_rmax=-1):
base = mol2_file.split('.')[0]
splat_range = 0
fraction = 0.9
default_dx = 0.5
models = mol2.read_Mol2_file( mol2_file )
if(len( models )== 0):
return None,None,None
## for testing purpose, here only the first model in the mol2 file is converted ##
this_molecule = models[0]
## models should be a list of molecules, same ligand at different conformations ##
all_atoms = this_molecule.atom_list
if(len( all_atoms )== 0):
return None,None,None
# predefine some arrays we will need
xyz = flex.vec3_double()
charges = flex.double()
# keep track of the atom types we have encountered
for atom in all_atoms:
xyz.append( ( atom.X, atom.Y, atom.Z ) )
charges.append( atom.Q )
if(xyz.size() == 0):
return None,None,None
if (uniform):
density=flex.double(xyz.size(),1.0)
else: # use charges
density=charges
voxel_obj = math.sphere_voxel(np,splat_range,uniform,fix_dx,external_rmax, default_dx, fraction,xyz,density)
np = voxel_obj.np()
rmax=voxel_obj.rmax()/fraction
#print base, "RMAX: ", voxel_obj.rmax()
#print time.time()
grid_obj = math.sphere_grid(np, nmax)
pdb_out = False
grid_obj.clean_space(voxel_obj, pdb_out)
#print time.time(), "END GRID CONST"
grid_obj.construct_space_sum()
#print time.time(), "SS CALC"
mom_obj = math.zernike_moments(grid_obj, nmax)
#print time.time(), "END MOM CALC"
moments = mom_obj.moments()
if(coef_out):
nn = mom_obj.fnn()
easy_pickle.dump(base+'.nlm.pickle', moments.coefs() )
easy_pickle.dump(base+'.nn.pickle', nn.coefs() )
#
####### The following will be optional ###
if(shift):
shift = [rmax, rmax, rmax]
centered_xyz = voxel_obj.xyz() + shift
out_mol2_filename =base+'_centered.mol2'
for a,xyz in zip( all_atoms, centered_xyz):
a.X = xyz[0]
a.Y = xyz[1]
a.Z = xyz[2]
mol2.write_mol2( this_molecule, out_mol2_filename )
original_map = voxel_obj.map()
xplor_map_type( original_map, np, rmax, file_name=base+'_pdb.xplor')
######## Calculate Intnesity Profile ############
if(calc_intensity):
q_array = flex.double( range(51) )/100.0
z_model = zm.zernike_model( moments, q_array, rmax, nmax)
nn = mom_obj.fnn()
intensity = z_model.calc_intensity(nn)
#iq_file = open(base+"_"+str(nmax)+"_"+str(int(rmax*fraction))+".zi", 'w')
iq_file = open(base+"_"+str(nmax)+".zi", 'w')
for qq, ii in zip(q_array, intensity):
print>>iq_file, qq,ii
iq_file.close()
####### END of Intensity Calculation ###########
if(buildmap):
print "grid setting up..."
zga = math.zernike_grid(np_on_grid, nmax, False)
print "grid setting up...Done"
zga.load_coefs(moments.nlm(), moments.coefs() )
print "start reconstruction"
map=flex.abs(zga.f() )
print "finished reconstruction"
xplor_map_type( map, np_on_grid, rmax, file_name=base+'.xplor')
ccp4_map_type( map, np_on_grid, rmax, file_name=base+'.ccp4' )
return mom_obj, voxel_obj, models
def build_map(nmax, shapes, coefs, codes, pdb_models):
np_on_grid = 30
zga = math.zernike_grid(np_on_grid, nmax, False)
ref_nlm_array = math.nlm_array(nmax)
mov_nlm_array = math.nlm_array(nmax)
nlm = ref_nlm_array.nlm()
nlm_total = ref_nlm_array.coefs().size()
top_cc = flex.double()
ntop = shapes.ntop
rank = 0
ave_c = flex.complex_double(nlm_total, 0)
if (pdb_models is not None):
ref_nlm_array.load_coefs(nlm, pdb_models[0].nlm_coef)
fraction = 0
else:
c = coefs[shapes.best_indices[0]][0:nlm_total]
ave_c = c.deep_copy()
ref_nlm_array.load_coefs(nlm, c)
rank = 1
filename = "m" + str(rank) + "_" + shapes.best_codes[0] + ".xplor"
fraction = write_xplor(zga, nlm, c, np_on_grid, shapes.best_rmax[0],
filename)
print rank, shapes.best_codes[0], fraction
ref_mean, ref_s = get_mean_sigma(ref_nlm_array)
mean_frac = fraction
mean_sqr_frac = fraction * fraction
for ii, code in zip(shapes.best_indices[rank:], shapes.best_codes[rank:]):
coef = coefs[ii][0:nlm_total]
mov_nlm_array.load_coefs(nlm, coef)
mov_mean, mov_s = get_mean_sigma(mov_nlm_array)
align_obj = fft_align.align(ref_nlm_array,
mov_nlm_array,
nmax=nmax,
refine=True)
new_c = align_obj.moving_nlm.coefs()
cc = align_obj.get_cc()
top_cc.append(cc)
ave_c = ave_c + new_c
filename = "m" + str(rank + 1) + "_" + code + ".xplor"
fraction = write_xplor(zga, nlm, new_c, np_on_grid,
shapes.best_rmax[rank], filename)
rank = rank + 1
print rank, code, fraction
mean_frac = mean_frac + fraction
mean_sqr_frac = mean_sqr_frac + fraction * fraction
#sphere_volume = 4.0/3.0*smath.pi*rmax*rmax*rmax
rmax = shapes.best_rmax[0]
sphere_volume = rmax * rmax * rmax * 8.0
mean_frac = mean_frac / ntop
sigma_frac = smath.sqrt(mean_sqr_frac / ntop - mean_frac * mean_frac)
print "Volume is ", mean_frac * sphere_volume, "+/-", sigma_frac * sphere_volume
#### Write average map ####
filename = "ave_map.xplor"
write_xplor(zga, nlm, ave_c / ntop, np_on_grid, rmax, filename)
return top_cc