def expand_crystal(structure, n=1, name='XTAL'):
"""Expands the contents of a structure to a crystal of a given size. Returns
a `` StructureHolder`` entity instance.
Arguments:
- structure: ``Structure`` entity instance.
- n: number number of unit-cell layers.
- name: optional name.
Requires a PDB file with correct CRYST1 field and space group information.
"""
sh = structure.header
sn = structure.name
fmx = sh['uc_fmx']
omx = sh['uc_omx']
# get initial coorinates
atoms = einput(structure, 'A')
coords = array([atoms.getData('coords')]) # fake 3D
# expand the coordinates to crystal
all_coords = coords_to_crystal(coords, fmx, omx, n)
structures = StructureHolder(name)
rng = range(-n, n + 1) # a range like -2, -1, 0, 1, 2
vectors = [(x, y, z) for x in rng for y in rng for z in rng]
for i, (u, v, w) in enumerate(vectors):
new_structure = copy(structure)
new_atoms = einput(new_structure, 'A')
new_coords = all_coords[i, 0]
for (atom_id, new_coord) in izip(atoms.keys(), new_coords):
new_atoms[atom_id].coords = new_coord
new_structure.setName("%s_%s%s%s" % (sn, u, v, w))
structures.addChild(new_structure)
return structures
def expand_crystal(structure, n=1, name='XTAL'):
"""Expands the contents of a structure to a crystal of a given size. Returns
a `` StructureHolder`` entity instance.
Arguments:
- structure: ``Structure`` entity instance.
- n: number number of unit-cell layers.
- name: optional name.
Requires a PDB file with correct CRYST1 field and space group information.
"""
sh = structure.header
sn = structure.name
fmx = sh['uc_fmx']
omx = sh['uc_omx']
# get initial coorinates
atoms = einput(structure, 'A')
coords = array([atoms.getData('coords')]) # fake 3D
# expand the coordinates to crystal
all_coords = coords_to_crystal(coords, fmx, omx, n)
structures = StructureHolder(name)
rng = range(-n, n + 1) # a range like -2, -1, 0, 1, 2
vectors = [(x, y, z) for x in rng for y in rng for z in rng]
for i, (u, v, w) in enumerate(vectors):
new_structure = copy(structure)
new_atoms = einput(new_structure, 'A')
new_coords = all_coords[i, 0]
for (atom_id, new_coord) in izip(atoms.keys(), new_coords):
new_atoms[atom_id].coords = new_coord
new_structure.setName("%s_%s%s%s" % (sn, u, v, w))
structures.addChild(new_structure)
return structures
def _prepare_asa(entities, symmetry_mode=None, crystal_mode=None, points=960, \
**kwargs):
"""Prepares the atomic solvent-accessible surface area (ASA) calculation.
Arguments:
- entities: input entities for ASA calculation (most commondly a
structure entity).
- symmetry_mode (str): One of 'uc', 'bio' or 'table'. This defines the
transformations of applied to the coordinates of the input entities.
It is one of 'bio', 'uc' or 'table'. Where 'bio' and 'uc' are
transformations to create the biological molecule or unit-cell from
the PDB header. The 'table' uses transformation matrices derived from
space-group information only using crystallographic tables(requires
``cctbx``).
- crystal_mode (int): Defines the number of unit-cells to expand the
initial unit-cell into. The number of unit-cells in each direction
i.e. 1 is makes a total of 27 unit cells: (-1, 0, 1) == 3, 3^3 == 27
- points: number of points on atom spheres higher is slower but more
accurate.
Additional keyworded arguments are passed to the ``_run_asa`` function.
"""
# generate uniform points on the unit-sphere
spoints = sphere_points(points)
# prepare entities for asa calculation
# free-floating area mode
result = {}
atoms = einput(entities, 'A')
if not symmetry_mode and not crystal_mode:
coords = array(atoms.getData('coords', forgiving=False))
coords = array([[coords]]) # fake 3D and 4D
idx_to_id = dict(enumerate(atoms.getData('getFull_id', \
forgiving=False, method=True)))
asas = _run_asa(atoms, coords, spoints, **kwargs)
for idx in xrange(asas.shape[0]):
result[idx_to_id[idx]] = asas[idx]
# crystal-contact area mode
elif symmetry_mode in ('table', 'uc'):
structure = einput(entities, 'S').values()[0]
sh = structure.header
coords = array(atoms.getData('coords', forgiving=False))
idx_to_id = dict(enumerate(atoms.getData('getFull_id', \
forgiving=False, method=True)))
# expand to unit-cell, real 3D
coords = coords_to_symmetry(coords, \
sh[symmetry_mode + '_fmx'], \
sh[symmetry_mode + '_omx'], \
sh[symmetry_mode + '_mxs'], \
symmetry_mode)
# expand to crystal, real 4D
if crystal_mode:
coords = coords_to_crystal(coords, \
sh[symmetry_mode + '_fmx'], \
sh[symmetry_mode + '_omx'], \
crystal_mode) # real 4D
else:
coords = array([coords]) # fake 4D
asas = _run_asa(atoms, coords, spoints, **kwargs)
for idx in xrange(asas.shape[0]):
result[idx_to_id[idx]] = asas[idx]
# biological area mode
elif symmetry_mode == 'bio':
structure = einput(entities, 'S').values()[0]
chains = einput(entities, 'C')
sh = structure.header
start = 0
for chain_ids, mx_num in sh['bio_cmx']:
sel = chains.selectChildren(chain_ids, 'contains', 'id').values()
atoms = einput(sel, 'A')
coords = array(atoms.getData('coords', forgiving=False))
idx_to_id = dict(enumerate(atoms.getData('getFull_id', \
forgiving=False, method=True)))
stop = start + mx_num
coords = coords_to_symmetry(coords, \
sh['uc_fmx'], \
sh['uc_omx'], \
sh['bio_mxs'][start:stop], \
symmetry_mode)
coords = array([coords])
start = stop
asas = _run_asa(atoms, coords, spoints, **kwargs)
for idx in xrange(asas.shape[0]):
result[idx_to_id[idx]] = asas[idx]
return result
def _prepare_asa(entities, symmetry_mode=None, crystal_mode=None, points=960, \
**kwargs):
"""Prepares the atomic solvent-accessible surface area (ASA) calculation.
Arguments:
- entities: input entities for ASA calculation (most commondly a
structure entity).
- symmetry_mode (str): One of 'uc', 'bio' or 'table'. This defines the
transformations of applied to the coordinates of the input entities.
It is one of 'bio', 'uc' or 'table'. Where 'bio' and 'uc' are
transformations to create the biological molecule or unit-cell from
the PDB header. The 'table' uses transformation matrices derived from
space-group information only using crystallographic tables(requires
``cctbx``).
- crystal_mode (int): Defines the number of unit-cells to expand the
initial unit-cell into. The number of unit-cells in each direction
i.e. 1 is makes a total of 27 unit cells: (-1, 0, 1) == 3, 3^3 == 27
- points: number of points on atom spheres higher is slower but more
accurate.
Additional keyworded arguments are passed to the ``_run_asa`` function.
"""
# generate uniform points on the unit-sphere
spoints = sphere_points(points)
# prepare entities for asa calculation
# free-floating area mode
result = {}
atoms = einput(entities, 'A')
if not symmetry_mode and not crystal_mode:
coords = array(atoms.getData('coords', forgiving=False))
coords = array([[coords]]) # fake 3D and 4D
idx_to_id = dict(enumerate(atoms.getData('getFull_id', \
forgiving=False, method=True)))
asas = _run_asa(atoms, coords, spoints, **kwargs)
for idx in xrange(asas.shape[0]):
result[idx_to_id[idx]] = asas[idx]
# crystal-contact area mode
elif symmetry_mode in ('table', 'uc'):
structure = einput(entities, 'S').values()[0]
sh = structure.header
coords = array(atoms.getData('coords', forgiving=False))
idx_to_id = dict(enumerate(atoms.getData('getFull_id', \
forgiving=False, method=True)))
# expand to unit-cell, real 3D
coords = coords_to_symmetry(coords, \
sh[symmetry_mode + '_fmx'], \
sh[symmetry_mode + '_omx'], \
sh[symmetry_mode + '_mxs'], \
symmetry_mode)
# expand to crystal, real 4D
if crystal_mode:
coords = coords_to_crystal(coords, \
sh[symmetry_mode + '_fmx'], \
sh[symmetry_mode + '_omx'], \
crystal_mode) # real 4D
else:
coords = array([coords]) # fake 4D
asas = _run_asa(atoms, coords, spoints, **kwargs)
for idx in xrange(asas.shape[0]):
result[idx_to_id[idx]] = asas[idx]
# biological area mode
elif symmetry_mode == 'bio':
structure = einput(entities, 'S').values()[0]
chains = einput(entities, 'C')
sh = structure.header
start = 0
for chain_ids, mx_num in sh['bio_cmx']:
sel = chains.selectChildren(chain_ids, 'contains', 'id').values()
atoms = einput(sel, 'A')
coords = array(atoms.getData('coords', forgiving=False))
idx_to_id = dict(enumerate(atoms.getData('getFull_id', \
forgiving=False, method=True)))
stop = start + mx_num
coords = coords_to_symmetry(coords, \
sh['uc_fmx'], \
sh['uc_omx'], \
sh['bio_mxs'][start:stop], \
symmetry_mode)
coords = array([coords])
start = stop
asas = _run_asa(atoms, coords, spoints, **kwargs)
for idx in xrange(asas.shape[0]):
result[idx_to_id[idx]] = asas[idx]
return result
def _prepare_contacts(query, model=None, level='A', search_limit=6.0, \
contact_mode='diff_chain', symmetry_mode=None, \
crystal_mode=None, **kwargs):
"""Prepares distance contact calculations.
Arguments:
- query(entitie[s]): query entitie[s] for contact calculation
(most commonly a structure entity).
- model(entity): a Model entity which will be transformed according to
symmetry_mode and crystal_mode. (most commonly it is the same as the
query)
- level(str): The level in the hierarchy at which distances will be
calculated (most commonly 'A' for atoms)
- search_limit(float): maximum distance in Angstrom's
- contact_mode(str): One of "diff_cell", "diff_sym", "diff_chain".
Defines the allowed contacts i.e. requires that contacts are by
entities, which have: "diff_cell" different unit cells; "diff_sym"
different symmetry operators (if in the same unit cell) "diff_chain"
with different chain ids (if in the same unit cell and symmetry).
- symmetry_mode (str): One of 'uc', 'bio' or 'table'. This defines the
transformations of applied to the coordinates of the input entities.
It is one of 'bio', 'uc' or 'table'. Where 'bio' and 'uc' are
transformations to create the biological molecule or unit-cell from
the PDB header. The 'table' uses transformation matrices derived from
space-group information only using crystallographic tables(requires
``cctbx``).
- crystal_mode (int): Defines the number of unit-cells to expand the
initial unit-cell into. The number of unit cells in each direction
i.e. 1 is makes a total of 27 unit cells: (-1, 0, 1) == 3, 3^3 == 27
Additional arguments are passed to the ``cnt_loop`` Cython function.
"""
contact_mode = {'diff_asu' :0,
'diff_sym' :1,
'diff_chain':2 }[contact_mode]
# determine unique structure
structure = einput(query, 'S').values()[0]
sh = structure.header
# if not specified otherwise the lattice is the first model
lattice = model or structure[(0,)]
lents = einput(lattice, level)
lents_ids = lents.getData('getFull_id', forgiving=False, method=True)
lcoords = array(lents.getData('coords', forgiving=False))
qents = einput(query, level)
qents_ids = qents.getData('getFull_id', forgiving=False, method=True)
qcoords = array(qents.getData('coords', forgiving=False))
if symmetry_mode:
if symmetry_mode == 'table':
lcoords = coords_to_symmetry(lcoords, \
sh['table_fmx'], \
sh['table_omx'], \
sh['table_mxs'], \
symmetry_mode)
elif symmetry_mode == 'uc':
lcoords = coords_to_symmetry(lcoords, \
sh['uc_fmx'], \
sh['uc_omx'], \
sh['uc_mxs'], \
symmetry_mode)
elif symmetry_mode == 'bio':
# TODO see asa
raise ValueError("Unsupported symmetry_mode: %s" % symmetry_mode)
else:
raise ValueError("Unsupported symmetry_mode: %s" % symmetry_mode)
else:
lcoords = array([lcoords]) # fake 3D
if crystal_mode:
zero_tra = {1:13, 2:62, 3:171}[crystal_mode]
# 0,0,0 translation is: Thickened cube numbers:
# a(n)=n*(n^2+(n-1)^2)+(n-1)*2*n*(n-1).
# 1, 14, 63, 172, 365, 666, 1099, 1688, 2457, 3430, 4631, 6084, 7813 ...
if symmetry_mode == 'table':
lcoords = coords_to_crystal(lcoords, \
sh['table_fmx'], \
sh['table_omx'], \
crystal_mode)
elif symmetry_mode == 'uc':
lcoords = coords_to_crystal(lcoords, \
sh['uc_fmx'], \
sh['uc_omx'], \
crystal_mode)
else:
raise ValueError('crystal_mode not possible for "bio" symmetry')
else:
zero_tra = 0
lcoords = array([lcoords]) # fake 4D
shape = lcoords.shape
lcoords = lcoords.reshape((shape[0] * shape[1] * shape[2], shape[3]))
box = r_[qcoords.min(axis=0) - search_limit, \
qcoords.max(axis=0) + search_limit]
lc = [] # lattice chain
qc = [] # query chain
lchains = [i[2] for i in lents_ids]
qchains = [i[2] for i in qents_ids]
allchains = set()
allchains.update(lchains)
allchains.update(qchains)
chain2id = dict(zip(allchains, range(len(allchains))))
for lent_id in lents_ids:
lc.append(chain2id[lent_id[2]])
for qent_id in qents_ids:
qc.append(chain2id[qent_id[2]])
lc = array(lc, dtype=int64)
qc = array(qc, dtype=int64)
# here we leave python
(idxc, n_src, n_asu, n_sym, n_tra, n_dst) = cnt_loop(\
qcoords, lcoords, qc, lc, shape[1], shape[2], \
zero_tra, contact_mode, search_limit, box, \
**kwargs)
result = defaultdict(dict)
for contact in xrange(idxc):
qent_id = qents_ids[n_src[contact]]
lent_id = lents_ids[n_asu[contact]]
result[qent_id][lent_id] = (sqrt(n_dst[contact]), n_tra[contact], n_sym[contact])
return result
def _prepare_contacts(query, model=None, level='A', search_limit=6.0, \
contact_mode='diff_chain', symmetry_mode=None, \
crystal_mode=None, **kwargs):
"""Prepares distance contact calculations.
Arguments:
- query(entitie[s]): query entitie[s] for contact calculation
(most commonly a structure entity).
- model(entity): a Model entity which will be transformed according to
symmetry_mode and crystal_mode. (most commonly it is the same as the
query)
- level(str): The level in the hierarchy at which distances will be
calculated (most commonly 'A' for atoms)
- search_limit(float): maximum distance in Angstrom's
- contact_mode(str): One of "diff_cell", "diff_sym", "diff_chain".
Defines the allowed contacts i.e. requires that contacts are by
entities, which have: "diff_cell" different unit cells; "diff_sym"
different symmetry operators (if in the same unit cell) "diff_chain"
with different chain ids (if in the same unit cell and symmetry).
- symmetry_mode (str): One of 'uc', 'bio' or 'table'. This defines the
transformations of applied to the coordinates of the input entities.
It is one of 'bio', 'uc' or 'table'. Where 'bio' and 'uc' are
transformations to create the biological molecule or unit-cell from
the PDB header. The 'table' uses transformation matrices derived from
space-group information only using crystallographic tables(requires
``cctbx``).
- crystal_mode (int): Defines the number of unit-cells to expand the
initial unit-cell into. The number of unit cells in each direction
i.e. 1 is makes a total of 27 unit cells: (-1, 0, 1) == 3, 3^3 == 27
Additional arguments are passed to the ``cnt_loop`` Cython function.
"""
contact_mode = {
'diff_asu': 0,
'diff_sym': 1,
'diff_chain': 2
}[contact_mode]
# determine unique structure
structure = einput(query, 'S').values()[0]
sh = structure.header
# if not specified otherwise the lattice is the first model
lattice = model or structure[(0, )]
lents = einput(lattice, level)
lents_ids = lents.getData('getFull_id', forgiving=False, method=True)
lcoords = array(lents.getData('coords', forgiving=False))
qents = einput(query, level)
qents_ids = qents.getData('getFull_id', forgiving=False, method=True)
qcoords = array(qents.getData('coords', forgiving=False))
if symmetry_mode:
if symmetry_mode == 'table':
lcoords = coords_to_symmetry(lcoords, \
sh['table_fmx'], \
sh['table_omx'], \
sh['table_mxs'], \
symmetry_mode)
elif symmetry_mode == 'uc':
lcoords = coords_to_symmetry(lcoords, \
sh['uc_fmx'], \
sh['uc_omx'], \
sh['uc_mxs'], \
symmetry_mode)
elif symmetry_mode == 'bio':
# TODO see asa
raise ValueError("Unsupported symmetry_mode: %s" % symmetry_mode)
else:
raise ValueError("Unsupported symmetry_mode: %s" % symmetry_mode)
else:
lcoords = array([lcoords]) # fake 3D
if crystal_mode:
zero_tra = {1: 13, 2: 62, 3: 171}[crystal_mode]
# 0,0,0 translation is: Thickened cube numbers:
# a(n)=n*(n^2+(n-1)^2)+(n-1)*2*n*(n-1).
# 1, 14, 63, 172, 365, 666, 1099, 1688, 2457, 3430, 4631, 6084, 7813 ...
if symmetry_mode == 'table':
lcoords = coords_to_crystal(lcoords, \
sh['table_fmx'], \
sh['table_omx'], \
crystal_mode)
elif symmetry_mode == 'uc':
lcoords = coords_to_crystal(lcoords, \
sh['uc_fmx'], \
sh['uc_omx'], \
crystal_mode)
else:
raise ValueError('crystal_mode not possible for "bio" symmetry')
else:
zero_tra = 0
lcoords = array([lcoords]) # fake 4D
shape = lcoords.shape
lcoords = lcoords.reshape((shape[0] * shape[1] * shape[2], shape[3]))
box = r_[qcoords.min(axis=0) - search_limit, \
qcoords.max(axis=0) + search_limit]
lc = [] # lattice chain
qc = [] # query chain
lchains = [i[2] for i in lents_ids]
qchains = [i[2] for i in qents_ids]
allchains = set()
allchains.update(lchains)
allchains.update(qchains)
chain2id = dict(zip(allchains, range(len(allchains))))
for lent_id in lents_ids:
lc.append(chain2id[lent_id[2]])
for qent_id in qents_ids:
qc.append(chain2id[qent_id[2]])
lc = array(lc, dtype=int64)
qc = array(qc, dtype=int64)
# here we leave python
(idxc, n_src, n_asu, n_sym, n_tra, n_dst) = cnt_loop(\
qcoords, lcoords, qc, lc, shape[1], shape[2], \
zero_tra, contact_mode, search_limit, box, \
**kwargs)
result = defaultdict(dict)
for contact in xrange(idxc):
qent_id = qents_ids[n_src[contact]]
lent_id = lents_ids[n_asu[contact]]
result[qent_id][lent_id] = (sqrt(n_dst[contact]), n_tra[contact],
n_sym[contact])
return result