def setWeights(self, weights):
"""Set atomic weights."""
if self._n_atoms == 0:
raise AttributeError('first set reference coordinates')
try:
self._weights = checkWeights(weights, self._n_atoms, None)
except ValueError:
weights = checkWeights(weights, self.numSelected(), None)
if not self._weights:
self._weights = ones((self._n_atoms, 1), dtype=float)
self._weights[self._indices, :] = weights
def setWeights(self, weights):
"""Set atomic weights."""
if self._n_atoms == 0:
raise AttributeError('first set reference coordinates')
try:
self._weights = checkWeights(weights, self._n_atoms, None)
except ValueError:
weights = checkWeights(weights, self.numSelected(), None)
if not self._weights:
self._weights = ones((self._n_atoms, 1), dtype=float)
self._weights[self._indices, :] = weights
def setWeights(self, weights):
"""Set atomic weights."""
if self._n_atoms == 0:
raise AttributeError('coordinates are not set')
try:
self._weights = checkWeights(weights, self._n_atoms, self._n_csets)
except ValueError:
weights = checkWeights(weights, self.numSelected(), self._n_csets)
if not self._weights:
self._weights = np.ones((self._n_csets, self._n_atoms, 1), dtype=float)
self._weights[self._indices, :] = weights
def setWeights(self, weights):
"""Set atomic weights."""
if self._n_atoms == 0:
raise AttributeError('coordinates are not set')
try:
self._weights = checkWeights(weights, self._n_atoms, self._n_csets)
except ValueError:
weights = checkWeights(weights, self.numSelected(), self._n_csets)
if not self._weights:
self._weights = np.ones((self._n_csets, self._n_atoms, 1), dtype=float)
self._weights[self._indices, :] = weights
def calcTransformation(mobile, target, weights=None):
"""Returns a :class:`Transformation` instance which, when applied to the
atoms in *mobile*, minimizes the weighted RMSD between *mobile* and
*target*. *mobile* and *target* may be NumPy coordinate arrays, or
:class:`.Atomic` instances, e.g. :class:`.AtomGroup`, :class:`.Chain`,
or :class:`.Selection`."""
if not isinstance(mobile, np.ndarray):
try:
mob = mobile._getCoords()
except AttributeError:
raise TypeError('mobile must be a numpy array or an object '
'with getCoords method')
else:
mob = mobile
if not isinstance(target, np.ndarray):
try:
tar = target._getCoords()
except AttributeError:
raise TypeError('target must be a numpy array or an object '
'with getCoords method')
else:
tar = target
if mob.shape != tar.shape:
raise ValueError('reference and target coordinate arrays '
'must have same number of atoms')
if mob.shape[1] != 3:
raise ValueError('reference and target must be coordinate arrays')
if weights is not None:
weights = checkWeights(weights, mob.shape[0])
return Transformation(*getTransformation(mob, tar, weights))
def setWeights(self, weights):
"""Set atomic weights."""
n_atoms = self.numAtoms()
n_modesets = self.numModeSets()
if n_atoms > 0 and n_modesets > 0:
self._weights = checkWeights(weights, n_atoms, n_modesets)
else:
raise RuntimeError('modesets must be set before weights can be assigned')
def addModeSet(self, modeset, weights=None, label=None):
"""Adds a modeset or modesets to the mode ensemble."""
if isinstance(modeset, (NMA, ModeSet, Mode)):
modesets = [modeset]
else:
modesets = modeset
if label is not None:
if np.isscalar(label):
labels = [label]
else:
labels = label
if len(labels) != len(modesets):
raise ValueError('labels should have the same length as modesets')
if not self._labels and labels:
self._labels = ['']*len(self._modesets)
self._labels.extend(labels)
if not labels and self._labels:
labels = ['']*len(modesets)
self._labels.extend(labels)
for i in range(len(modesets)):
modeset = modesets[i]
if isinstance(modeset, NMA):
modeset = modeset[:]
elif isinstance(modeset, Mode):
modeset = ModeSet(modeset.getModel(), [modeset.getIndex()])
if not isinstance(modeset, ModeSet):
raise TypeError('modesets should be a list of Mode or ModeSet instances')
if self._modesets:
if modeset.numAtoms() != self.numAtoms():
raise ValueError('to be added, modesets should contain exactly %d atoms'%self.numAtoms())
if modeset.numModes() < self.numModes():
raise ValueError('to be added, modesets should contain at least %d modes'%self.numModes())
if modeset.numModes() > self.numModes():
modeset = modeset[:self.numModes()]
self._modesets.append(modeset)
else:
self._modesets = [modeset]
if weights is None:
weights = np.ones((len(modesets), self.numAtoms(), 1))
if self._weights is not None:
self._weights = np.concatenate((self._weights, weights), axis=0)
else:
self._weights = weights
else:
weights = checkWeights(weights, self.numAtoms(), len(modesets))
if self._weights is None:
self._weights = np.ones((self.numModeSets()-len(modesets), self.numAtoms(), 1))
self._weights = np.concatenate((self._weights, weights), axis=0)
def calcRMSD(reference, target=None, weights=None):
"""Returns root-mean-square deviation (RMSD) between reference and target
coordinates.
.. ipython:: python
ens = loadEnsemble('p38_X-ray.ens.npz')
ens.getRMSDs()
calcRMSD(ens)
calcRMSD(ens.getCoords(), ens.getCoordsets(), ens.getWeights())"""
if isinstance(reference, np.ndarray):
ref = reference
else:
try:
ref = reference._getCoords()
except AttributeError:
raise TypeError('reference must be a numpy array or an object '
'with getCoords method')
if target is None:
try:
target = reference._getCoordsets()
except AttributeError:
pass
if weights is None:
try:
weights = reference._getWeights()
except AttributeError:
pass
if ref.ndim != 2 or ref.shape[1] != 3:
raise ValueError('reference must have shape (n_atoms, 3)')
if isinstance(target, np.ndarray):
tar = target
else:
try:
tar = target._getCoords()
except AttributeError:
raise TypeError('target must be a numpy array or an object '
'with getCoords method')
if tar.ndim not in (2, 3) or tar.shape[-1] != 3:
raise ValueError('target must have shape ([n_confs,] n_atoms, 3)')
if ref.shape != tar.shape[-2:]:
raise ValueError('reference and target arrays must have the same '
'number of atoms')
if weights is not None:
n_atoms = len(ref)
n_csets = 1 if tar.ndim == 2 else len(tar)
weights = checkWeights(weights, n_atoms, n_csets)
return getRMSD(ref, tar, weights)
def addCoordset(self, coords, weights=None, label=None, **kwargs):
"""Add coordinate set(s) to the ensemble. *coords* must be a Numpy
array with suitable shape and dimensionality, or an object with
:meth:`getCoordsets`. *weights* is an optional argument.
If provided, its length must match number of atoms. Weights of
missing (not resolved) atoms must be ``0`` and weights of those
that are resolved can be anything greater than ``0``. If not
provided, weights of all atoms for this coordinate set will be
set equal to ``1``. *label*, which may be a PDB identifier or a
list of identifiers, is used to label conformations."""
degeneracy = kwargs.pop('degeneracy', False)
adddata = kwargs.pop('data', None)
atoms = coords
n_atoms = self._n_atoms
n_select = self.numSelected()
n_confs = self.numCoordsets()
try:
if degeneracy:
if self._coords is not None:
if isinstance(coords, Ensemble):
coords = coords._getCoords(selected=False)
elif hasattr(coords, '_getCoords'):
coords = coords._getCoords()
else:
if isinstance(coords, Ensemble):
coords = coords.getCoords(selected=False)
elif hasattr(coords, 'getCoords'):
coords = coords.getCoords()
else:
if self._coords is not None:
if isinstance(coords, Ensemble):
coords = coords._getCoordsets(selected=False)
elif hasattr(coords, '_getCoordsets'):
coords = coords._getCoordsets()
else:
if isinstance(coords, Ensemble):
coords = coords.getCoordsets(selected=False)
elif hasattr(coords, 'getCoordsets'):
coords = coords.getCoordsets()
except AttributeError:
label = label or 'Unknown'
else:
if coords is None:
raise ValueError('coordinates are not set')
elif label is None and isinstance(atoms, Atomic):
if not isinstance(atoms, AtomGroup):
ag = atoms.getAtomGroup()
else:
ag = atoms
label = ag.getTitle()
if coords.shape[0] < ag.numCoordsets():
label += '_m' + str(atoms.getACSIndex())
else:
label = label or 'Unknown'
# check coordinates
try:
checkCoords(coords, csets=True, natoms=n_atoms)
except:
try:
checkCoords(coords, csets=True, natoms=n_select)
except TypeError:
raise TypeError('coords must be a numpy array or an object '
'with `getCoords` method')
if coords.ndim == 2:
n_nodes, _ = coords.shape
coords = coords.reshape((1, n_nodes, 3))
n_csets = 1
else:
n_csets, n_nodes, _ = coords.shape
if degeneracy:
coords = coords[:1]
n_repeats = 1 if degeneracy else n_csets
if not n_atoms:
self._n_atoms = n_nodes
n_atoms = self._n_atoms
if n_nodes == n_select and self.isSelected():
full_coords = np.repeat(self._coords[np.newaxis, :, :],
n_csets,
axis=0)
full_coords[:, self._indices, :] = coords
coords = full_coords
# check weights
if weights is None:
weights = np.ones((n_csets, n_atoms, 1), dtype=float)
else:
weights = checkWeights(weights, n_atoms, n_csets)
if degeneracy:
weights = weights[:1]
# check sequences
seqs = None
sequence = kwargs.pop('sequence', None)
if hasattr(atoms, 'getSequence'):
if sequence is not None:
LOGGER.warn(
'sequence is supplied though coords has getSequence')
sequence = atoms.getSequence()
seqs = [sequence for _ in range(n_repeats)]
else:
if sequence is None:
try:
sequence = self._atoms.getSequence()
except AttributeError:
if self._msa:
sequence = ''.join('X' for _ in range(n_atoms))
# sequence and seqs remains to be None if MSA has not been created
if isinstance(sequence, Sequence):
seqs = [str(sequence)]
elif isinstance(sequence, MSA):
seqs = [str(seq) for seq in sequence]
elif np.isscalar(sequence):
seqs = [sequence for _ in range(n_repeats)]
if seqs:
if len(seqs) != n_repeats:
raise ValueError(
'the number of sequences should be either one or '
'that of coordsets')
# assign new values
# update labels
if n_csets > 1 and not degeneracy:
if isinstance(label, str):
labels = [
'{0}_m{1}'.format(label, i + 1) for i in range(n_csets)
]
else:
if len(label) != n_csets:
raise ValueError('length of label and number of '
'coordinate sets must be the same')
labels = label
else:
labels = [label] if np.isscalar(label) else label
self._labels.extend(labels)
# update sequences
if seqs:
msa = MSA(seqs, title=self.getTitle(), labels=labels)
if self._msa is None:
if n_confs > 0:
def_seqs = np.chararray((n_confs, n_atoms))
def_seqs[:] = 'X'
old_labels = [self._labels[i] for i in range(n_confs)]
self._msa = MSA(def_seqs,
title=self.getTitle(),
labels=old_labels)
self._msa.extend(msa)
else:
self._msa = msa
else:
self._msa.extend(msa)
# update coordinates
if self._confs is None and self._weights is None:
self._confs = coords
self._weights = weights
elif self._confs is not None and self._weights is not None:
self._confs = np.concatenate((self._confs, coords), axis=0)
self._weights = np.concatenate((self._weights, weights), axis=0)
else:
raise RuntimeError('_confs and _weights must be set or None at '
'the same time')
# appending new data
if self._data is not None and adddata is not None:
if self._data is None:
self._data = {}
if adddata is None:
adddata = {}
all_keys = set(list(self._data.keys()) + list(adddata.keys()))
for key in all_keys:
if key in self._data:
data = self._data[key]
if key not in adddata:
shape = [n_repeats]
for s in data.shape[1:]:
shape.append(s)
newdata = np.zeros(shape, dtype=data.dtype)
else:
newdata = np.asarray(adddata[key])
if newdata.shape[0] != n_repeats:
raise ValueError(
'the length of data["%s"] does not match that of coords'
% key)
else:
newdata = np.asarray(adddata[key])
shape = [self._n_csets]
for s in newdata.shape[1:]:
shape.append(s)
data = np.zeros(shape, dtype=newdata.dtype)
self._data[key] = np.concatenate((data, newdata), axis=0)
# update the number of coordinate sets
self._n_csets += n_repeats
def setWeights(self, weights):
"""Set atomic weights."""
if self._n_atoms == 0:
raise AttributeError('coordinates must be set first')
self._weights = checkWeights(weights, self._n_atoms, None)
def addCoordset(self, coords, weights=None, label=None, degeneracy=False):
"""Add coordinate set(s) to the ensemble. *coords* must be a Numpy
array with suitable shape and dimensionality, or an object with
:meth:`getCoordsets` method. *weights* is an optional argument.
If provided, its length must match number of atoms. Weights of
missing (not resolved) atoms must be ``0`` and weights of those
that are resolved can be anything greater than ``0``. If not
provided, weights of all atoms for this coordinate set will be
set equal to ``1``. *label*, which may be a PDB identifier or a
list of identifiers, is used to label conformations."""
atoms = coords
try:
if self._coords is not None:
if isinstance(coords, Ensemble):
coords = coords._getCoordsets(selected=False)
elif hasattr(coords, '_getCoordsets'):
coords = coords._getCoordsets()
else:
if isinstance(coords, Ensemble):
coords = coords.getCoordsets(selected=False)
elif hasattr(coords, 'getCoordsets'):
coords = coords.getCoordsets()
except AttributeError:
label = label or 'Unknown'
else:
if coords is None:
raise ValueError('coordinates are not set')
elif label is None and isinstance(atoms, Atomic):
ag = atoms
if not isinstance(atoms, AtomGroup):
ag = atoms.getAtomGroup()
label = ag.getTitle()
if coords.shape[0] < ag.numCoordsets():
label += 'm' + str(atoms.getACSIndex())
else:
label = label or str(coords)
try:
checkCoords(coords, csets=True, natoms=self._n_atoms)
except TypeError:
raise TypeError('coords must be a Numpy array or must have '
'`getCoords` attribute')
if coords.ndim == 2:
coords = coords.reshape((1, self._n_atoms, 3))
n_csets, n_atoms, _ = coords.shape
if not self._n_atoms:
self._n_atoms = n_atoms
if weights is None:
weights = np.ones((n_csets, n_atoms, 1), dtype=float)
else:
weights = checkWeights(weights, n_atoms, n_csets)
if n_csets > 1:
if degeneracy == False:
if isinstance(label, str):
self._labels.extend('{0}_m{1}'.format(label, i + 1)
for i in range(n_csets))
else:
if len(label) != n_csets:
raise ValueError('length of label and number of '
'coordinate sets must be the same')
self._labels.extend(label)
else:
self._labels.append(label)
coords = np.reshape(
coords[0], (1, coords[0].shape[0], coords[0].shape[1]))
weights = np.reshape(
weights[0], (1, weights[0].shape[0], weights[0].shape[1]))
else:
self._labels.append(label)
if self._confs is None and self._weights is None:
self._confs = coords
self._weights = weights
if degeneracy == False:
self._n_csets = n_csets
else:
self._n_csets = 1
elif self._confs is not None and self._weights is not None:
self._confs = np.concatenate((self._confs, coords), axis=0)
self._weights = np.concatenate((self._weights, weights), axis=0)
if degeneracy == False:
self._n_csets += n_csets
else:
self._n_csets += 1
else:
raise RuntimeError('_confs and _weights must be set or None at '
'the same time')
def setWeights(self, weights):
"""Set atomic weights."""
if self._n_atoms == 0:
raise AttributeError('first set reference coordinates')
self._weights = checkWeights(weights, self._n_atoms, None)
def setWeights(self, weights):
"""Set atomic weights."""
if self._n_atoms == 0:
raise AttributeError('coordinates must be set first')
self._weights = checkWeights(weights, self._n_atoms, None)
def addCoordset(self, coords, weights=None, label=None, **kwargs):
"""Add coordinate set(s) to the ensemble. *coords* must be a Numpy
array with suitable shape and dimensionality, or an object with
:meth:`getCoordsets`. *weights* is an optional argument.
If provided, its length must match number of atoms. Weights of
missing (not resolved) atoms must be ``0`` and weights of those
that are resolved can be anything greater than ``0``. If not
provided, weights of all atoms for this coordinate set will be
set equal to ``1``. *label*, which may be a PDB identifier or a
list of identifiers, is used to label conformations."""
degeneracy = kwargs.pop('degeneracy', False)
atoms = coords
n_atoms = self._n_atoms
n_select = self.numSelected()
n_confs = self.numCoordsets()
try:
if degeneracy:
if self._coords is not None:
if isinstance(coords, Ensemble):
coords = coords._getCoords(selected=False)
elif hasattr(coords, '_getCoords'):
coords = coords._getCoords()
else:
if isinstance(coords, Ensemble):
coords = coords.getCoords(selected=False)
elif hasattr(coords, 'getCoords'):
coords = coords.getCoords()
else:
if self._coords is not None:
if isinstance(coords, Ensemble):
coords = coords._getCoordsets(selected=False)
elif hasattr(coords, '_getCoordsets'):
coords = coords._getCoordsets()
else:
if isinstance(coords, Ensemble):
coords = coords.getCoordsets(selected=False)
elif hasattr(coords, 'getCoordsets'):
coords = coords.getCoordsets()
except AttributeError:
label = label or 'Unknown'
else:
if coords is None:
raise ValueError('coordinates are not set')
elif label is None and isinstance(atoms, Atomic):
if not isinstance(atoms, AtomGroup):
ag = atoms.getAtomGroup()
else:
ag = atoms
label = ag.getTitle()
if coords.shape[0] < ag.numCoordsets():
label += '_m' + str(atoms.getACSIndex())
else:
label = label or 'Unknown'
# check coordinates
try:
checkCoords(coords, csets=True, natoms=n_atoms)
except:
try:
checkCoords(coords, csets=True, natoms=n_select)
except TypeError:
raise TypeError('coords must be a numpy array or an object '
'with `getCoords` method')
if coords.ndim == 2:
n_nodes, _ = coords.shape
coords = coords.reshape((1, n_nodes, 3))
n_csets = 1
else:
n_csets, n_nodes, _ = coords.shape
if degeneracy:
coords = coords[:1]
n_repeats = 1 if degeneracy else n_csets
if not n_atoms:
self._n_atoms = n_nodes
if n_nodes == n_select and self.isSelected():
full_coords = np.repeat(self._coords[np.newaxis, :, :], n_csets, axis=0)
full_coords[:, self._indices, :] = coords
coords = full_coords
# check weights
if weights is None:
weights = np.ones((n_csets, n_atoms, 1), dtype=float)
else:
weights = checkWeights(weights, n_atoms, n_csets)
if degeneracy:
weights = weights[:1]
# check sequences
seqs = None
sequence = kwargs.pop('sequence', None)
if hasattr(atoms, 'getSequence'):
if sequence is not None:
LOGGER.warn('sequence is supplied though coords has getSequence')
sequence = atoms.getSequence()
seqs = [sequence for _ in range(n_repeats)]
else:
if sequence is None:
try:
sequence = self.getAtoms().getSequence()
except AttributeError:
if self._msa:
sequence = ''.join('X' for _ in range(n_atoms))
# sequence and seqs remains to be None if MSA has not been created
if isinstance(sequence, Sequence):
seqs = [str(sequence)]
elif isinstance(sequence, MSA):
seqs = [str(seq) for seq in sequence]
elif np.isscalar(sequence):
seqs = [sequence for _ in range(n_repeats)]
if seqs:
if len(seqs) != n_repeats:
raise ValueError('the number of sequences should be either one or '
'that of coordsets')
# assign new values
# update labels
if n_csets > 1 and not degeneracy:
if isinstance(label, str):
labels = ['{0}_m{1}'.format(label, i+1) for i in range(n_csets)]
else:
if len(label) != n_csets:
raise ValueError('length of label and number of '
'coordinate sets must be the same')
labels = label
else:
labels = [label] if np.isscalar(label) else label
self._labels.extend(labels)
# update sequences
if seqs:
msa = MSA(seqs, title=self.getTitle(), labels=labels)
if self._msa is None:
if n_confs > 0:
def_seqs = np.chararray((n_confs, n_atoms))
def_seqs[:] = 'X'
old_labels = [self._labels[i] for i in range(n_confs)]
self._msa = MSA(def_seqs, title=self.getTitle(), labels=old_labels)
self._msa.extend(msa)
else:
self._msa = msa
else:
self._msa.extend(msa)
# update coordinates
if self._confs is None and self._weights is None:
self._confs = coords
self._weights = weights
self._n_csets = n_repeats
elif self._confs is not None and self._weights is not None:
self._confs = np.concatenate((self._confs, coords), axis=0)
self._weights = np.concatenate((self._weights, weights), axis=0)
self._n_csets += n_repeats
else:
raise RuntimeError('_confs and _weights must be set or None at '
'the same time')