def _reduceModel(matrix, system):
"""This is the underlying function that reduces models, which shall
remain private. ``system`` is a boolean array where **True** indicates
system nodes."""
linalg = importLA()
other = np.invert(system)
ss = matrix[system, :][:, system]
so = matrix[system, :][:, other]
os = matrix[other, :][:, system]
oo = matrix[other, :][:, other]
if other.any():
try:
invoo = linalg.inv(oo)
except:
invoo = linalg.pinv(oo)
matrix = ss - np.dot(so, np.dot(invoo, os))
else:
matrix = ss
return matrix
def getTransformation(mob, tar, weights=None):
linalg = importLA()
if weights is None:
mob_com = mob.mean(0)
tar_com = tar.mean(0)
mob = mob - mob_com
tar = tar - tar_com
matrix = np.dot(tar.T, mob)
else:
weights_sum = weights.sum()
weights_dot = np.dot(weights.T, weights)
mob_com = (mob * weights).sum(axis=0) / weights_sum
tar_com = (tar * weights).sum(axis=0) / weights_sum
mob = mob - mob_com
tar = tar - tar_com
matrix = np.dot((tar * weights).T, (mob * weights)) / weights_dot
U, s, Vh = linalg.svd(matrix)
Id = np.array([ [1, 0, 0],
[0, 1, 0],
[0, 0, np.sign(linalg.det(matrix))] ])
rotation = np.dot(Vh.T, np.dot(Id, U.T))
return rotation, tar_com - np.dot(mob_com, rotation)
def _getEigvecs(modes, row_norm=False):
if isinstance(modes, (ModeSet, NMA)):
V = modes.getEigvecs()
elif isinstance(modes, Mode):
V = modes.getEigvec()
elif isinstance(modes, np.ndarray):
V = modes
else:
try:
mode0 = modes[0]
if isinstance(mode0, Mode):
V = np.empty((len(mode0), 0))
for mode in modes:
assert isinstance(mode,
Mode), 'Modes should be a list of modes.'
v = mode.getEigvec()
v = np.expand_dims(v, axis=1)
V = np.hstack((V, v))
else:
V = np.array(modes)
except TypeError:
TypeError('Modes should be a list of modes.')
if V.ndim == 1:
V = np.expand_dims(V, axis=1)
# normalize the rows so that feature vectors are unit vectors
if row_norm:
la = importLA()
norms = la.norm(V, axis=1)
N = np.diag(div0(1., norms))
V = np.dot(N, V)
return V
def calcProjection(ensemble, modes, rmsd=True, norm=True):
"""Returns projection of conformational deviations onto given modes.
*ensemble* coordinates are used to calculate the deviations that are
projected onto *modes*. For K conformations and M modes, a (K,M)
matrix is returned.
:arg ensemble: an ensemble, trajectory or a conformation for which
deviation(s) will be projected, or a deformation vector
:type ensemble: :class:`.Ensemble`, :class:`.Conformation`,
:class:`.Vector`, :class:`.Trajectory`
:arg modes: up to three normal modes
:type modes: :class:`.Mode`, :class:`.ModeSet`, :class:`.NMA`
By default root-mean-square deviation (RMSD) along the normal mode is
calculated. To calculate the projection pass ``rmsd=True``.
:class:`.Vector` instances are accepted as *ensemble* argument to allow
for projecting a deformation vector onto normal modes."""
if not isinstance(ensemble, (Ensemble, Conformation, Vector, TrajBase)):
raise TypeError('ensemble must be Ensemble, Conformation, Vector, '
'or a TrajBase, not {0}'.format(type(ensemble)))
if not isinstance(modes, (NMA, ModeSet, VectorBase)):
raise TypeError('rows must be NMA, ModeSet, or Mode, not {0}'.format(
type(modes)))
if not modes.is3d():
raise ValueError('modes must be 3-dimensional')
if isinstance(ensemble, Vector):
n_atoms = ensemble.numAtoms()
else:
n_atoms = ensemble.numSelected()
if n_atoms != modes.numAtoms():
raise ValueError('number of atoms are not the same')
if isinstance(ensemble, Vector):
if not ensemble.is3d():
raise ValueError('ensemble must be a 3d vector instance')
deviations = ensemble._getArray()
elif isinstance(ensemble, (Ensemble, Conformation)):
deviations = ensemble.getDeviations()
else:
nfi = ensemble.nextIndex()
ensemble.goto(0)
deviations = np.array([frame.getDeviations() for frame in ensemble])
ensemble.goto(nfi)
if deviations.ndim == 3:
deviations = deviations.reshape(
(deviations.shape[0], deviations.shape[1] * 3))
elif deviations.ndim == 2:
deviations = deviations.reshape((1, deviations.shape[0] * 3))
else:
deviations = deviations.reshape((1, deviations.shape[0]))
la = importLA()
if norm:
N = la.norm(deviations)
if N != 0:
deviations = deviations / N
projection = np.dot(deviations, modes._getArray())
if rmsd:
projection = (1 / (n_atoms**0.5)) * projection
return projection
def calcProjection(ensemble, modes, rmsd=True, norm=True):
"""Returns projection of conformational deviations onto given modes.
*ensemble* coordinates are used to calculate the deviations that are
projected onto *modes*. For K conformations and M modes, a (K,M)
matrix is returned.
:arg ensemble: an ensemble, trajectory or a conformation for which
deviation(s) will be projected, or a deformation vector
:type ensemble: :class:`.Ensemble`, :class:`.Conformation`,
:class:`.Vector`, :class:`.Trajectory`
:arg modes: up to three normal modes
:type modes: :class:`.Mode`, :class:`.ModeSet`, :class:`.NMA`
By default root-mean-square deviation (RMSD) along the normal mode is
calculated. To calculate the projection pass ``rmsd=True``.
:class:`.Vector` instances are accepted as *ensemble* argument to allow
for projecting a deformation vector onto normal modes."""
if not isinstance(ensemble, (Ensemble, Conformation, Vector, TrajBase)):
raise TypeError('ensemble must be Ensemble, Conformation, Vector, '
'or a TrajBase, not {0}'.format(type(ensemble)))
if not isinstance(modes, (NMA, ModeSet, VectorBase)):
raise TypeError('rows must be NMA, ModeSet, or Mode, not {0}'
.format(type(modes)))
if not modes.is3d():
raise ValueError('modes must be 3-dimensional')
if isinstance(ensemble, Vector):
n_atoms = ensemble.numAtoms()
else:
n_atoms = ensemble.numSelected()
if n_atoms != modes.numAtoms():
raise ValueError('number of atoms are not the same')
if isinstance(ensemble, Vector):
if not ensemble.is3d():
raise ValueError('ensemble must be a 3d vector instance')
deviations = ensemble._getArray()
elif isinstance(ensemble, (Ensemble, Conformation)):
deviations = ensemble.getDeviations()
else:
nfi = ensemble.nextIndex()
ensemble.goto(0)
deviations = np.array([frame.getDeviations() for frame in ensemble])
ensemble.goto(nfi)
if deviations.ndim == 3:
deviations = deviations.reshape((deviations.shape[0],
deviations.shape[1] * 3))
elif deviations.ndim == 2:
deviations = deviations.reshape((1, deviations.shape[0] * 3))
else:
deviations = deviations.reshape((1, deviations.shape[0]))
la = importLA()
if norm:
N = la.norm(deviations)
if N != 0:
deviations = deviations / N
projection = np.dot(deviations, modes._getArray())
if rmsd:
projection = (1 / (n_atoms ** 0.5)) * projection
return projection
def calcHitTime(model, method='standard'):
"""Returns the hit and commute times between pairs of nodes calculated
based on a :class:`.NMA` object.
.. [CB95] Chennubhotla C., Bahar I. Signal Propagation in Proteins and Relation
to Equilibrium Fluctuations. *PLoS Comput Biol* **2007** 3(9).
:arg model: model to be used to calculate hit times
:type model: :class:`.NMA`
:arg method: method to be used to calculate hit times. Available options are
``"standard"`` or ``"kirchhoff"``. Default is ``"standard"``
:type method: str
:returns: (:class:`~numpy.ndarray`, :class:`~numpy.ndarray`)
"""
try:
K = model.getKirchhoff()
except AttributeError:
raise TypeError('model must be an NMA instance')
if K is None:
raise ValueError('model not built')
method = method.lower()
D = np.diag(K)
A = np.diag(D) - K
start = time.time()
linalg = importLA()
if method == 'standard':
st = D / sum(D)
P = np.dot(np.diag(D**(-1)), A)
W = np.ones((len(st), 1)) * st.T
Z = linalg.pinv(np.eye(P.shape[0], P.shape[1]) - P + W)
H = np.ones((len(st), 1)) * np.diag(Z).T - Z
H = H / W
H = H.T
elif method == 'kirchhoff':
K_inv = linalg.pinv(K)
sum_D = sum(D)
T1 = (sum_D * np.ones((len(D),1)) * np.diag(K_inv)).T
T2 = sum_D * K_inv
T3_i = np.dot((np.ones((len(D),1)) * D), K_inv)
H = T1 - T2 + T3_i - T3_i.T
C = H + H.T
LOGGER.debug('Hit and commute times are calculated in {0:.2f}s.'
.format(time.time()-start))
return H, C
def _superpose(self, **kwargs):
"""Superpose conformations and update coordinates."""
indices = self._indices
weights = self._weights
mobs = self._confs
if indices is None:
idx = False
tar = self._coords
movs = None
else:
idx = True
if self._weights is not None:
weights = weights[indices]
tar = self._coords[indices]
movs = self._confs
linalg = importLA()
svd = linalg.svd
det = linalg.det
if weights is None:
tar_com = tar.mean(0)
tar_org = (tar - tar_com)
mob_org = zeros(tar_org.shape, dtype=mobs.dtype)
tar_org = tar_org.T
else:
weights_sum = weights.sum()
weights_dot = dot(weights.T, weights)
tar_com = (tar * weights).sum(axis=0) / weights_sum
tar_org = (tar - tar_com)
mob_org = zeros(tar_org.shape, dtype=mobs.dtype)
LOGGER.progress('Superposing ', len(mobs), '_prody_ensemble')
for i, mob in enumerate(mobs):
if idx:
mob = mob[indices]
if weights is None:
mob_com = mob.mean(0)
matrix = dot(tar_org, subtract(mob, mob_com, mob_org))
else:
mob_com = (mob * weights).sum(axis=0) / weights_sum
subtract(mob, mob_com, mob_org)
matrix = dot((tar_org * weights).T,
(mob_org * weights)) / weights_dot
U, s, Vh = svd(matrix)
Id = array([[1, 0, 0], [0, 1, 0], [0, 0, sign(det(matrix))]])
rotation = dot(Vh.T, dot(Id, U.T))
if movs is None:
mobs[i] = dot(mob_org, rotation)
add(mobs[i], tar_com, mobs[i])
else:
add(dot(movs[i], rotation),
(tar_com - dot(mob_com, rotation)), movs[i])
LOGGER.update(i, '_prody_ensemble')
LOGGER.clear()
def _superpose(self, **kwargs):
"""Superpose conformations and update coordinates."""
indices = self._indices
weights = self._weights
mobs = self._confs
if indices is None:
idx = False
tar = self._coords
movs = None
else:
idx = True
if self._weights is not None:
weights = weights[indices]
tar = self._coords[indices]
movs = self._confs
linalg = importLA()
svd = linalg.svd
det = linalg.det
if weights is None:
tar_com = tar.mean(0)
tar_org = (tar - tar_com)
mob_org = zeros(tar_org.shape, dtype=mobs.dtype)
tar_org = tar_org.T
else:
weights_sum = weights.sum()
weights_dot = dot(weights.T, weights)
tar_com = (tar * weights).sum(axis=0) / weights_sum
tar_org = (tar - tar_com)
mob_org = zeros(tar_org.shape, dtype=mobs.dtype)
LOGGER.progress('Superposing ', len(mobs), '_prody_ensemble')
for i, mob in enumerate(mobs):
if idx:
mob = mob[indices]
if weights is None:
mob_com = mob.mean(0)
matrix = dot(tar_org, subtract(mob, mob_com, mob_org))
else:
mob_com = (mob * weights).sum(axis=0) / weights_sum
subtract(mob, mob_com, mob_org)
matrix = dot((tar_org * weights).T,
(mob_org * weights)) / weights_dot
U, s, Vh = svd(matrix)
Id = array([[1, 0, 0], [0, 1, 0], [0, 0, sign(det(matrix))]])
rotation = dot(Vh.T, dot(Id, U.T))
if movs is None:
mobs[i] = dot(mob_org, rotation)
add(mobs[i], tar_com, mobs[i])
else:
add(dot(movs[i], rotation),
(tar_com - dot(mob_com, rotation)), movs[i])
LOGGER.update(i + 1, label='_prody_ensemble')
LOGGER.finish()
def performSVD(self, coordsets):
"""Calculate principal modes using singular value decomposition (SVD).
*coordsets* argument may be a :class:`.Atomic`, :class:`.Ensemble`,
or :class:`numpy.ndarray` instance. If *coordsets* is a numpy array,
its shape must be ``(n_csets, n_atoms, 3)``. Note that coordinate
sets must be aligned prior to SVD calculations.
This is a considerably faster way of performing PCA calculations
compared to eigenvalue decomposition of covariance matrix, but is
an approximate method when heterogeneous datasets are analyzed.
Covariance method should be preferred over this one for analysis of
ensembles with missing atomic data. See :ref:`pca-xray-calculations`
example for comparison of results from SVD and covariance methods."""
linalg = importLA()
start = time.time()
if not isinstance(coordsets, (Ensemble, Atomic, np.ndarray)):
raise TypeError('coordsets must be an Ensemble, Atomic, Numpy '
'array instance')
if isinstance(coordsets, np.ndarray):
if (coordsets.ndim != 3 or coordsets.shape[2] != 3
or coordsets.dtype not in (np.float32, float)):
raise ValueError('coordsets is not a valid coordinate array')
deviations = coordsets - coordsets.mean(0)
else:
if isinstance(coordsets, Ensemble):
deviations = coordsets.getDeviations()
elif isinstance(coordsets, Atomic):
deviations = (coordsets._getCoordsets() -
coordsets._getCoords())
n_confs = deviations.shape[0]
if n_confs < 3:
raise ValueError('coordsets must have more than 3 coordinate sets')
n_atoms = deviations.shape[1]
if n_atoms < 3:
raise ValueError('coordsets must have more than 3 atoms')
dof = n_atoms * 3
deviations = deviations.reshape((n_confs, dof)).T
vectors, values, self._temp = linalg.svd(deviations,
full_matrices=False)
values = (values**2) / n_confs
self._dof = dof
self._n_atoms = n_atoms
which = values > 1e-18
self._eigvals = values[which]
self._array = vectors[:, which]
self._vars = self._eigvals
self._trace = self._vars.sum()
self._n_modes = len(self._eigvals)
LOGGER.debug('{0} modes were calculated in {1:.2f}s.'.format(
self._n_modes,
time.time() - start))
def performSVD(self, coordsets):
"""Calculate principal modes using singular value decomposition (SVD).
*coordsets* argument may be a :class:`.Atomic`, :class:`.Ensemble`,
or :class:`numpy.ndarray` instance. If *coordsets* is a numpy array,
its shape must be ``(n_csets, n_atoms, 3)``. Note that coordinate
sets must be aligned prior to SVD calculations.
This is a considerably faster way of performing PCA calculations
compared to eigenvalue decomposition of covariance matrix, but is
an approximate method when heterogeneous datasets are analyzed.
Covariance method should be preferred over this one for analysis of
ensembles with missing atomic data. See :ref:`pca-xray-calculations`
example for comparison of results from SVD and covariance methods."""
linalg = importLA()
start = time.time()
if not isinstance(coordsets, (Ensemble, Atomic, np.ndarray)):
raise TypeError('coordsets must be an Ensemble, Atomic, Numpy '
'array instance')
if isinstance(coordsets, np.ndarray):
if (coordsets.ndim != 3 or coordsets.shape[2] != 3 or
coordsets.dtype not in (np.float32, float)):
raise ValueError('coordsets is not a valid coordinate array')
deviations = coordsets - coordsets.mean(0)
else:
if isinstance(coordsets, Ensemble):
deviations = coordsets.getDeviations()
elif isinstance(coordsets, Atomic):
deviations = (coordsets._getCoordsets() -
coordsets._getCoords())
n_confs = deviations.shape[0]
if n_confs < 3:
raise ValueError('coordsets must have more than 3 coordinate sets')
n_atoms = deviations.shape[1]
if n_atoms < 3:
raise ValueError('coordsets must have more than 3 atoms')
dof = n_atoms * 3
deviations = deviations.reshape((n_confs, dof)).T
vectors, values, self._temp = linalg.svd(deviations,
full_matrices=False)
values = (values ** 2) / n_confs
self._dof = dof
self._n_atoms = n_atoms
which = values > 1e-18
self._eigvals = values[which]
self._array = vectors[:, which]
self._vars = self._eigvals
self._trace = self._vars.sum()
self._n_modes = len(self._eigvals)
LOGGER.debug('{0} modes were calculated in {1:.2f}s.'
.format(self._n_modes, time.time()-start))
def _getEigvecs(modes, row_norm=False, dummy_mode=False):
la = importLA()
if isinstance(modes, (Mode, ModeSet, NMA)):
model = modes._model
if isinstance(model, MaskedGNM):
masked = model.masked
model.masked = True
V = modes.getArray()
model.masked = masked
else:
V = modes.getArray()
elif isinstance(modes, np.ndarray):
V = modes
else:
try:
mode0 = modes[0]
if isinstance(mode0, Mode):
V = np.empty((len(mode0), 0))
for mode in modes:
assert isinstance(mode,
Mode), 'Modes should be a list of modes.'
v = mode.getEigvec()
v = np.expand_dims(v, axis=1)
V = np.hstack((V, v))
else:
V = np.array(modes)
except TypeError:
raise TypeError('Modes should be a list of modes.')
if V.ndim == 1:
V = np.expand_dims(V, axis=1)
# add a dummy zero mode to the modeset
if dummy_mode:
v0 = V[:, 0]
if np.allclose(v0, np.mean(v0)):
dummy_mode = False
LOGGER.warn(
'at least one zero mode is detected therefore dummy mode will NOT be added'
)
if dummy_mode:
n, _ = V.shape
v0 = np.ones((n, 1), dtype=V.dtype)
v0 /= la.norm(v0)
V = np.hstack((v0, V))
LOGGER.debug('a dummy zero mode is added')
# normalize the rows so that feature vectors are unit vectors
if row_norm:
norms = la.norm(V, axis=1)
N = np.diag(div0(1., norms))
V = np.dot(N, V)
return V
def calcModes(self, n_modes=20, turbo=True):
"""Calculate principal (or essential) modes. This method uses
:func:`scipy.linalg.eigh`, or :func:`numpy.linalg.eigh`, function
to diagonalize the covariance matrix.
:arg n_modes: number of non-zero eigenvalues/vectors to calculate,
default is 20,
if **None** or ``'all'`` is given, all modes will be calculated
:type n_modes: int
:arg turbo: when available, use a memory intensive but faster way to
calculate modes, default is **True**
:type turbo: bool"""
linalg = importLA()
if self._cov is None:
raise ValueError('covariance matrix is not built or set')
start = time.time()
dof = self._dof
self._clear()
if str(n_modes).lower() == 'all':
n_modes = None
if linalg.__package__.startswith('scipy'):
if n_modes is None:
eigvals = None
n_modes = dof
else:
n_modes = int(n_modes)
if n_modes >= self._dof:
eigvals = None
n_modes = dof
else:
eigvals = (dof - n_modes, dof - 1)
values, vectors = linalg.eigh(self._cov,
turbo=turbo,
eigvals=eigvals)
else:
if n_modes is not None:
LOGGER.info('Scipy is not found, all modes are calculated.')
values, vectors = linalg.eigh(self._cov)
# Order by descending SV
revert = list(range(len(values) - 1, -1, -1))
values = values[revert]
vectors = vectors[:, revert]
which = values > 1e-8
self._eigvals = values[which]
self._array = vectors[:, which]
self._vars = self._eigvals
self._n_modes = len(self._eigvals)
LOGGER.debug('{0} modes were calculated in {1:.2f}s.'.format(
self._n_modes,
time.time() - start))
def superpose(self):
"""Superpose frame onto the trajectory reference coordinates. Note
that transformation matrix is calculated based on selected atoms and
applied to all atoms. If atom weights for the trajectory are set, they
will be used to calculate the transformation."""
traj = self._traj
indices = traj._indices
ag = traj._ag
if ag is None:
mob = mov = self._coords
else:
mob = mov = ag._getCoords()
weights = traj._weights
if indices is None:
tar = traj._coords
mov = None
else:
tar = traj._coords[indices]
mob = mob[indices]
if weights is not None:
weights = weights[indices]
linalg = importLA()
if weights is None:
mob_com = mob.mean(0)
mob_org = mob - mob_com
tar_com = tar.mean(0)
tar_org = tar - tar_com
matrix = np.dot(tar_org.T, mob_org)
else:
weights_sum = weights.sum()
weights_dot = np.dot(weights.T, weights)
mob_com = (mob * weights).sum(axis=0) / weights_sum
mob_org = mob - mob_com
tar_com = (tar * weights).sum(axis=0) / weights_sum
tar_org = tar - tar_com
matrix = np.dot((tar_org * weights).T,
(mob_org * weights)) / weights_dot
U, s, Vh = linalg.svd(matrix)
Id = np.array([ [1, 0, 0],
[0, 1, 0],
[0, 0, np.sign(linalg.det(matrix))] ])
rotation = np.dot(Vh.T, np.dot(Id, U.T))
if mov is None:
np.add(np.dot(mob_org, rotation), tar_com, mob)
else:
np.add(np.dot(mov, rotation),
(tar_com - np.dot(mob_com, rotation)), mov)
def calcEntropyTransfer(model, ind1, ind2, tau):
"""This function calculates the entropy transfer from residue indice
ind1 to ind2 for a given time constant tau based on GNM.
"""
if not isinstance(model, NMA):
raise TypeError('model must be a NMA instance')
elif model.is3d():
raise TypeError('model must be a 1-dimensional NMA instance')
linalg = importLA()
n_atoms = model.numAtoms()
n_modes = model.numModes()
eigvecs = model.getEigvecs().T
eigvals = model.getEigvals()
tau_0 = 1
T = 0
dummy1 = 0
dummy2 = 0
for k in range(n_modes):
dummy1 += 1.0 / eigvals[k] * eigvecs[k,ind2] * eigvecs[k,ind2]
dummy2 += 1.0 / eigvals[k] * eigvecs[k,ind2] * eigvecs[k,ind2] * np.exp(-eigvals[k]*tau/tau_0)
T += 0.5 * np.log(dummy1**2 - dummy2**2)
dummy1 = 0
dummy2 = 0
dummy3 = 0
dummy4 = 0
dummy5 = 0
dummy6 = 0
dummy7 = 0
dummy8 = 0
dummy9 = 0
dummy10 = 0
for k in range(n_modes):
dummy1 += 1.0 / eigvals[k] * eigvecs[k,ind1] * eigvecs[k,ind1]
dummy2 += 1.0 / eigvals[k] * eigvecs[k,ind2] * eigvecs[k,ind2]
dummy3 += 1.0 / eigvals[k] * eigvecs[k,ind1] * eigvecs[k,ind2]
dummy4 += 1.0 / eigvals[k] * eigvecs[k,ind2] * eigvecs[k,ind2] * np.exp(-eigvals[k]*tau/tau_0)
dummy5 += 1.0 / eigvals[k] * eigvecs[k,ind1] * eigvecs[k,ind2] * np.exp(-eigvals[k]*tau/tau_0)
dummy9 += 1.0 / eigvals[k] * eigvecs[k,ind2] * eigvecs[k,ind2] * np.exp(-eigvals[k]*tau/tau_0)
dummy6 = dummy5
dummy7 = dummy3
dummy8 = dummy2
dummy10 = dummy1
T -= 0.5 * np.log(dummy1*dummy2**2+2*dummy3*dummy4*dummy5-(dummy6**2+dummy7**2)*dummy8-dummy9**2*dummy10)
T -= 0.5 * np.log(dummy2)
T += 0.5 * np.log(dummy1*dummy2-dummy3**2)
return T
def calcOverallNetEntropyTransfer(model, turbo=False):
"""This function calculates the net entropy transfer for a whole structure
with a given time constant tau based on GNM.
"""
if not isinstance(model, NMA):
raise TypeError('model must be a NMA instance')
elif model.is3d():
raise TypeError('model must be a 1-dimensional NMA instance')
linalg = importLA()
n_atoms = model.numAtoms()
n_modes = model.numModes()
tau_max = 5.0
tau_step = 0.1
taus = np.arange(start=tau_step, stop=tau_max+1e-6, step=tau_step)
taus = np.insert(taus,0,0.000001)
numTaus = len(taus)
netEntropyTransfer = np.zeros((numTaus,n_atoms,n_atoms))
if turbo:
try:
from joblib import Parallel, delayed
import multiprocessing as mp
except:
LOGGER.report('joblib and multiprocessing is not imported. Running' +
'with sequential execution.')
LOGGER.timeit('_ent_trans')
n_cpu = mp.cpu_count()
netEntropyTransfer = Parallel(n_jobs=n_cpu)(delayed(calcAllEntropyTransfer)(model,taus[i]) \
for i in range(numTaus))
netEntropyTransfer = np.asarray(netEntropyTransfer)
else:
LOGGER.timeit('_ent_trans')
for i in range(len(taus)):
netEntropyTransfer[i,:,:] = calcAllEntropyTransfer(model,taus[i])
LOGGER.report('Net Entropy Transfer calculation is completed in %.1fs.',
'_ent_trans')
overallNetEntropyTransfer = np.zeros((n_atoms,n_atoms))
LOGGER.timeit('_num_int')
for i in range(n_atoms):
for j in range(n_atoms):
if i != j:
overallNetEntropyTransfer[i,j] = np.trapz(netEntropyTransfer[:,i,j],taus)
LOGGER.report('Numerical integration is completed in %.1fs.',
'_num_int')
return overallNetEntropyTransfer
def getNormDistFluct(self, coords):
"""Normalized distance fluctuation
"""
model = self.getModel()
LOGGER.info('Number of chains: {0}, chains: {1}.'
.format(len(list(set(coords.getChids()))), \
list(set(coords.getChids()))))
try:
#coords = coords.select('protein and name CA')
coords = (coords._getCoords() if hasattr(coords, '_getCoords') else
coords.getCoords())
except AttributeError:
try:
checkCoords(coords)
except TypeError:
raise TypeError('coords must be a Numpy array or an object '
'with `getCoords` method')
if not isinstance(model, NMA):
LOGGER.info('Calculating new model')
model = GNM('prot analysis')
model.buildKirchhoff(coords)
model.calcModes()
linalg = importLA()
n_atoms = model.numAtoms()
n_modes = model.numModes()
LOGGER.timeit('_ndf')
from .analysis import calcCrossCorr
from numpy import linalg as LA
# <dRi, dRi>, <dRj, dRj> = 1
crossC = 2-2*calcCrossCorr(model)
r_ij = np.zeros((n_atoms,n_atoms,3))
for i in range(n_atoms):
for j in range(i+1,n_atoms):
r_ij[i][j] = coords[j,:] - coords[i,:]
r_ij[j][i] = r_ij[i][j]
r_ij_n = LA.norm(r_ij, axis=2)
#with np.errstate(divide='ignore'):
r_ij_n[np.diag_indices_from(r_ij_n)] = 1e-5 # div by 0
crossC=abs(crossC)
normdistfluct = np.divide(np.sqrt(crossC),r_ij_n)
LOGGER.report('NDF calculated in %.2lfs.', label='_ndf')
normdistfluct[np.diag_indices_from(normdistfluct)] = 0 # div by 0
return normdistfluct
def calcModes(self, n_modes=20, turbo=True):
"""Calculate principal (or essential) modes. This method uses
:func:`scipy.linalg.eigh`, or :func:`numpy.linalg.eigh`, function
to diagonalize the covariance matrix.
:arg n_modes: number of non-zero eigenvalues/vectors to calculate,
default is 20,
if **None** or ``'all'`` is given, all modes will be calculated
:type n_modes: int
:arg turbo: when available, use a memory intensive but faster way to
calculate modes, default is **True**
:type turbo: bool"""
linalg = importLA()
if self._cov is None:
raise ValueError('covariance matrix is not built or set')
start = time.time()
dof = self._dof
self._clear()
if str(n_modes).lower() == 'all':
n_modes = None
if linalg.__package__.startswith('scipy'):
if n_modes is None:
eigvals = None
n_modes = dof
else:
n_modes = int(n_modes)
if n_modes >= self._dof:
eigvals = None
n_modes = dof
else:
eigvals = (dof - n_modes, dof - 1)
values, vectors = linalg.eigh(self._cov, turbo=turbo,
eigvals=eigvals)
else:
if n_modes is not None:
LOGGER.info('Scipy is not found, all modes are calculated.')
values, vectors = linalg.eigh(self._cov)
# Order by descending SV
revert = list(range(len(values)-1, -1, -1))
values = values[revert]
vectors = vectors[:, revert]
which = values > 1e-8
self._eigvals = values[which]
self._array = vectors[:, which]
self._vars = self._eigvals
self._n_modes = len(self._eigvals)
LOGGER.debug('{0} modes were calculated in {1:.2f}s.'
.format(self._n_modes, time.time()-start))
def getNormDistFluct(self, coords):
"""Normalized distance fluctuation
"""
model = self.getModel()
LOGGER.info('Number of chains: {0}, chains: {1}.'
.format(len(list(set(coords.getChids()))), \
list(set(coords.getChids()))))
try:
#coords = coords.select('protein and name CA')
coords = (coords._getCoords()
if hasattr(coords, '_getCoords') else coords.getCoords())
except AttributeError:
try:
checkCoords(coords)
except TypeError:
raise TypeError('coords must be a Numpy array or an object '
'with `getCoords` method')
if not isinstance(model, NMA):
LOGGER.info('Calculating new model')
model = GNM('prot analysis')
model.buildKirchhoff(coords)
model.calcModes()
linalg = importLA()
n_atoms = model.numAtoms()
n_modes = model.numModes()
LOGGER.timeit('_ndf')
from .analysis import calcCrossCorr
from numpy import linalg as LA
# <dRi, dRi>, <dRj, dRj> = 1
crossC = 2 - 2 * calcCrossCorr(model)
r_ij = np.zeros((n_atoms, n_atoms, 3))
for i in range(n_atoms):
for j in range(i + 1, n_atoms):
r_ij[i][j] = coords[j, :] - coords[i, :]
r_ij[j][i] = r_ij[i][j]
r_ij_n = LA.norm(r_ij, axis=2)
#with np.errstate(divide='ignore'):
r_ij_n[np.diag_indices_from(r_ij_n)] = 1e-5 # div by 0
crossC = abs(crossC)
normdistfluct = np.divide(np.sqrt(crossC), r_ij_n)
LOGGER.report('NDF calculated in %.2lfs.', label='_ndf')
normdistfluct[np.diag_indices_from(normdistfluct)] = 0 # div by 0
return normdistfluct
def SCN(M, **kwargs):
la = importLA()
total_count = kwargs.pop('total_count', None)
max_loops = kwargs.pop('max_loops', 100)
tol = kwargs.pop('tol', 1e-5)
N = M.copy()
n = 0
d0 = None
p = 1
last_p = None
while True:
C = np.diag(div0(1., np.sum(N, axis=0)))
N = np.dot(N, C)
R = np.diag(div0(1., np.sum(N, axis=1)))
N = np.dot(R, N)
n += 1
# check convergence of symmetry
d = np.mean(np.abs(N - N.T))
if d0 is not None:
p = div0(d, d0)
dp = np.abs(p - last_p)
if dp < tol:
break
else:
d0 = d
LOGGER.debug('Iteration {0}: d = {1}, p = {2}'.format(
str(n), str(d), str(p)))
last_p = p
if max_loops is not None:
if n >= max_loops:
LOGGER.warn('The SCN algorithm did not converge after {0} '
'iterations.'.format(max_loops))
break
# guarantee symmetry
N = (N + N.T) / 2.
if total_count is 'original':
total_count = np.sum(M)
if total_count is not None:
sum_N = np.sum(N)
k = total_count / sum_N
N = N * k
return N
def SCN(M, **kwargs):
la = importLA()
total_count = kwargs.pop('total_count', None)
max_loops = kwargs.pop('max_loops', 100)
tol = kwargs.pop('tol', 1e-5)
N = M.copy()
n = 0
d0 = None
p = 1
last_p = None
while True:
C = np.diag(div0(1., np.sum(N, axis=0)))
N = np.dot(N, C)
R = np.diag(div0(1., np.sum(N, axis=1)))
N = np.dot(R, N)
n += 1
# check convergence of symmetry
d = np.mean(np.abs(N - N.T))
if d0 is not None:
p = div0(d, d0)
dp = np.abs(p - last_p)
if dp < tol:
break
else:
d0 = d
LOGGER.debug('Iteration {0}: d = {1}, p = {2}'.format(str(n), str(d), str(p)))
last_p = p
if max_loops is not None:
if n >= max_loops:
LOGGER.warn('The SCN algorithm did not converge after {0} '
'iterations.'.format(max_loops))
break
# guarantee symmetry
N = (N + N.T) / 2.
if total_count is 'original':
total_count = np.sum(M)
if total_count is not None:
sum_N = np.sum(N)
k = total_count / sum_N
N = N * k
return N
def superpose(self):
"""Superpose frame onto the trajectory reference coordinates. Note
that transformation matrix is calculated based on selected atoms and
applied to all atoms. If atom weights for the trajectory are set, they
will be used to calculate the transformation."""
traj = self._traj
indices = traj._indices
mob = mov = self._getxyz()
weights = traj._weights
if indices is None:
tar = traj._coords
mov = None
else:
tar = traj._coords[indices]
mob = mob[indices]
if weights is not None:
weights = weights[indices]
linalg = importLA()
if weights is None:
mob_com = mob.mean(0)
mob_org = mob - mob_com
tar_com = tar.mean(0)
tar_org = tar - tar_com
matrix = np.dot(tar_org.T, mob_org)
else:
weights_sum = weights.sum()
weights_dot = np.dot(weights.T, weights)
mob_com = (mob * weights).sum(axis=0) / weights_sum
mob_org = mob - mob_com
tar_com = (tar * weights).sum(axis=0) / weights_sum
tar_org = tar - tar_com
matrix = np.dot((tar_org * weights).T,
(mob_org * weights)) / weights_dot
U, s, Vh = linalg.svd(matrix)
Id = np.array([[1, 0, 0], [0, 1, 0],
[0, 0, np.sign(linalg.det(matrix))]])
rotation = np.dot(Vh.T, np.dot(Id, U.T))
if mov is None:
np.add(np.dot(mob_org, rotation), tar_com, mob)
else:
np.add(np.dot(mov, rotation),
(tar_com - np.dot(mob_com, rotation)), mov)
def calcHitTime(self, method='Z'):
if self._affinity is None:
self._buildAffinity()
start = time.time()
linalg = importLA()
if method == 'Z':
D = self._diagonal
A = self._affinity
st = D / sum(D)
P = np.dot(np.diag(D**(-1)), A)
W = np.ones((len(st),1)) * st.T
Z = linalg.pinv(np.eye(P.shape[0], P.shape[1]) - P + W)
H = np.ones((len(st),1)) * np.diag(Z).T - Z
H = H / W
H = H.T
elif method == 'K':
K = self._kirchhoff
D = self._diagonal
K_inv = linalg.pinv(K)
sum_D = sum(D)
T1 = (sum_D * np.ones((len(D),1)) * np.diag(K_inv)).T
T2 = sum_D * K_inv
T3_i = np.dot((np.ones((len(D),1)) * D), K_inv)
H = T1 - T2 + T3_i - T3_i.T
self._hitTime = H
self._commuteTime = H + H.T
LOGGER.debug('Hitting and commute time are calculated in {0:.2f}s.'
.format(time.time()-start))
def calcHitTime(self, method='Z'):
if self._affinity is None:
self._buildAffinity()
start = time.time()
linalg = importLA()
if method == 'Z':
D = self._diagonal
A = self._affinity
st = D / sum(D)
P = np.dot(np.diag(D**(-1)), A)
W = np.ones((len(st),1)) * st.T
Z = linalg.pinv(np.eye(P.shape[0], P.shape[1]) - P + W)
H = np.ones((len(st),1)) * np.diag(Z).T - Z
H = H / W
H = H.T
elif method == 'K':
K = self._kirchhoff
D = self._diagonal
K_inv = linalg.pinv(K)
sum_D = sum(D)
T1 = (sum_D * np.ones((len(D),1)) * np.diag(K_inv)).T
T2 = sum_D * K_inv
T3_i = np.dot((np.ones((len(D),1)) * D), K_inv)
H = T1 - T2 + T3_i - T3_i.T
self._hitTime = H
self._commuteTime = H + H.T
LOGGER.debug('Hitting and commute time are calculated in {0:.2f}s.'
.format(time.time()-start))
def calcAllEntropyTransfer(model, tau):
"""This function calculates the net entropy transfer for a whole structure
with a given time constant tau based on GNM.
"""
if not isinstance(model, NMA):
raise TypeError('model must be a NMA instance')
elif model.is3d():
raise TypeError('model must be a 1-dimensional NMA instance')
linalg = importLA()
n_atoms = model.numAtoms()
n_modes = model.numModes()
entropyTransfer = np.zeros((n_atoms,n_atoms))
for i in range(n_atoms):
for j in range(n_atoms):
if i != j:
entropyTransfer[i,j]=calcEntropyTransfer(model,i,j,tau)
return entropyTransfer
def _getEigvecs(modes, row_norm=False, remove_zero_rows=False):
if isinstance(modes, (ModeSet, NMA)):
V = modes.getEigvecs()
elif isinstance(modes, Mode):
V = modes.getEigvec()
elif isinstance(modes, np.ndarray):
V = modes
else:
try:
mode0 = modes[0]
if isinstance(mode0, Mode):
V = np.empty((len(mode0),0))
for mode in modes:
assert isinstance(mode, Mode), 'Modes should be a list of modes.'
v = mode.getEigvec()
v = np.expand_dims(v, axis=1)
V = np.hstack((V, v))
else:
V = np.array(modes)
except TypeError:
raise TypeError('Modes should be a list of modes.')
if V.ndim == 1:
V = np.expand_dims(V, axis=1)
# normalize the rows so that feature vectors are unit vectors
if row_norm:
la = importLA()
norms = la.norm(V, axis=1)
N = np.diag(div0(1., norms))
V = np.dot(N, V)
# remove rows with all zeros
m, _ = V.shape
mask = np.ones(m, dtype=bool)
if remove_zero_rows:
mask = V.any(axis=1)
V = V[mask]
return V, mask
def _getEigvecs(modes, row_norm=False, remove_zero_rows=False):
if isinstance(modes, (ModeSet, NMA)):
V = modes.getEigvecs()
elif isinstance(modes, Mode):
V = modes.getEigvec()
elif isinstance(modes, np.ndarray):
V = modes
else:
try:
mode0 = modes[0]
if isinstance(mode0, Mode):
V = np.empty((len(mode0),0))
for mode in modes:
assert isinstance(mode, Mode), 'Modes should be a list of modes.'
v = mode.getEigvec()
v = np.expand_dims(v, axis=1)
V = np.hstack((V, v))
else:
V = np.array(modes)
except TypeError:
TypeError('Modes should be a list of modes.')
if V.ndim == 1:
V = np.expand_dims(V, axis=1)
# normalize the rows so that feature vectors are unit vectors
if row_norm:
la = importLA()
norms = la.norm(V, axis=1)
N = np.diag(div0(1., norms))
V = np.dot(N, V)
# remove rows with all zeros
m, _ = V.shape
mask = np.ones(m, dtype=bool)
if remove_zero_rows:
mask = V.any(axis=1)
V = V[mask]
return V, mask
def calcADPs(atom):
"""Calculate anisotropic displacement parameters (ADPs) from
anisotropic temperature factors (ATFs).
*atom* must have ATF values set for ADP calculation. ADPs are returned
as a tuple, i.e. (eigenvalues, eigenvectors)."""
linalg = importLA()
if not isinstance(atom, Atom):
raise TypeError('atom must be of type Atom, not {0:s}'
.format(type(atom)))
anisou = atom.getAnisou()
if anisou is None:
raise ValueError('atom does not have anisotropic temperature factors')
element = zeros((3,3))
element[0,0] = anisou[0]
element[1,1] = anisou[1]
element[2,2] = anisou[2]
element[0,1] = element[1,0] = anisou[3]
element[0,2] = element[2,0] = anisou[4]
element[1,2] = element[2,1] = anisou[5]
vals, vecs = linalg.eigh(element)
return vals[[2,1,0]], vecs[:, [2,1,0]]
def calcADPs(atom):
"""Calculate anisotropic displacement parameters (ADPs) from
anisotropic temperature factors (ATFs).
*atom* must have ATF values set for ADP calculation. ADPs are returned
as a tuple, i.e. (eigenvalues, eigenvectors)."""
linalg = importLA()
if not isinstance(atom, Atom):
raise TypeError('atom must be of type Atom, not {0}'.format(
type(atom)))
anisou = atom.getAnisou()
if anisou is None:
raise ValueError('atom does not have anisotropic temperature '
'factors')
element = zeros((3, 3))
element[0, 0] = anisou[0]
element[1, 1] = anisou[1]
element[2, 2] = anisou[2]
element[0, 1] = element[1, 0] = anisou[3]
element[0, 2] = element[2, 0] = anisou[4]
element[1, 2] = element[2, 1] = anisou[5]
vals, vecs = linalg.eigh(element)
return vals[[2, 1, 0]], vecs[:, [2, 1, 0]]
# -*- coding: utf-8 -*-
"""This module defines a class and a function for explicit membrane ANM calculations."""
import numpy as np
from prody import LOGGER, PY3K
from prody.atomic import Atomic, AtomGroup
from prody.utilities import importLA, checkCoords, copy
from numpy import sqrt, zeros, array, ceil, dot
from .anm import ANM
from .gnm import checkENMParameters
from .editing import _reduceModel
LA = importLA()
inv = LA.inv
pinv = LA.pinv
norm = LA.norm
__all__ = ['exANM']
class exANM(ANM):
"""Class for explicit ANM (exANM) method ([FT00]_).
Optional arguments build a membrane lattice permit analysis of membrane
effect on elastic network models in *exANM* method described in [TL12]_.
.. [TL12] Lezon TR, Bahar I, Constraints Imposed by the Membrane
Selectively Guide the Alternating Access Dynamics of the Glutamate
Transporter GltPh
def calcModes(self, n_modes=20, zeros=False, turbo=True):
"""Calculate normal modes. This method uses :func:`scipy.linalg.eigh`
function to diagonalize the Hessian matrix. When Scipy is not found,
:func:`numpy.linalg.eigh` is used.
:arg n_modes: number of non-zero eigenvalues/vectors to calculate.
If ``None`` is given, all modes will be calculated.
:type n_modes: int or None, default is 20
:arg zeros: If ``True``, modes with zero eigenvalues will be kept.
:type zeros: bool, default is ``False``
:arg turbo: Use a memory intensive, but faster way to calculate modes.
:type turbo: bool, default is ``True``
"""
if self._hessian is None:
raise ValueError('Hessian matrix is not built or set')
assert n_modes is None or isinstance(n_modes, int) and n_modes > 0, \
'n_modes must be a positive integer'
assert isinstance(zeros, bool), 'zeros must be a boolean'
assert isinstance(turbo, bool), 'turbo must be a boolean'
linalg = importLA()
LOGGER.timeit('_anm_calc_modes')
shift = 5
if linalg.__package__.startswith('scipy'):
if n_modes is None:
eigvals = None
n_modes = self._dof
else:
if n_modes >= self._dof:
eigvals = None
n_modes = self._dof
else:
eigvals = (0, n_modes + shift)
if eigvals:
turbo = False
if isinstance(self._hessian, np.ndarray):
values, vectors = linalg.eigh(self._hessian, turbo=turbo,
eigvals=eigvals)
else:
try:
from scipy.sparse import linalg as scipy_sparse_la
except ImportError:
raise ImportError('failed to import scipy.sparse.linalg, '
'which is required for sparse matrix '
'decomposition')
try:
values, vectors = (
scipy_sparse_la.eigsh(self._hessian, k=n_modes+6,
which='SA'))
except:
values, vectors = (
scipy_sparse_la.eigen_symmetric(self._hessian,
k=n_modes+6,
which='SA'))
else:
if n_modes is not None:
LOGGER.info('Scipy is not found, all modes are calculated.')
values, vectors = linalg.eigh(self._hessian)
n_zeros = sum(values < ZERO)
if n_zeros < 6:
LOGGER.warning('Less than 6 zero eigenvalues are calculated.')
shift = n_zeros - 1
elif n_zeros > 6:
LOGGER.warning('More than 6 zero eigenvalues are calculated.')
shift = n_zeros - 1
if zeros:
shift = -1
self._eigvals = values[1+shift:]
self._vars = 1 / self._eigvals
self._trace = self._vars.sum()
if shift:
self._array = vectors[:, 1+shift:].copy()
else:
self._array = vectors
self._n_modes = len(self._eigvals)
LOGGER.report('{0} modes were calculated in %.2fs.'
.format(self._n_modes), label='_anm_calc_modes')
def reduceModel(model, atoms, select):
"""Return reduced NMA model. Reduces a :class:`.NMA` model to a subset of
*atoms* matching *select*. This function behaves differently depending on
the type of the *model* argument. For :class:`.ANM` and :class:`.GNM` or
other :class:`.NMA` models, force constant matrix for system of interest
(specified by the *select*) is derived from the force constant matrix for
the *model* by assuming that for any given displacement of the system of
interest, other atoms move along in such a way as to minimize the potential
energy. This is based on the formulation in [KH00]_. For :class:`.PCA`
models, this function simply takes the sub-covariance matrix for selection.
.. [KH00] Konrad H, Andrei-Jose P, Serge D, Marie-Claire BF, Gerald RK.
Harmonicity in slow protein dynamics. *Chem Phys* **2000** 261:25-37.
:arg model: dynamics model
:type model: :class:`.ANM`, :class:`.GNM`, or :class:`.PCA`
:arg atoms: atoms that were used to build the model
:type atoms: :class:`.Atomic`
:arg select: an atom selection or a selection string
:type select: :class:`.Selection`, str
:returns: (:class:`.NMA`, :class:`.Selection`)"""
linalg = importLA()
if not isinstance(model, NMA):
raise TypeError('model must be an NMA instance, not {0}'
.format(type(model)))
if not isinstance(atoms, (AtomGroup, AtomSubset, AtomMap)):
raise TypeError('atoms type is not valid')
if len(atoms) <= 1:
raise TypeError('atoms must contain more than 1 atoms')
if isinstance(model, GNM):
matrix = model._kirchhoff
elif isinstance(model, ANM):
matrix = model._hessian
elif isinstance(model, PCA):
matrix = model._cov
else:
raise TypeError('model does not have a valid type derived from NMA')
if matrix is None:
raise ValueError('model matrix (Hessian/Kirchhoff/Covariance) is not '
'built')
if isinstance(select, str):
system = SELECT.getBoolArray(atoms, select)
n_sel = sum(system)
if n_sel == 0:
raise ValueError('select matches 0 atoms')
if len(atoms) == n_sel:
raise ValueError('select matches all atoms')
if isinstance(atoms, AtomGroup):
ag = atoms
which = np.arange(len(atoms))[system]
else:
ag = atoms.getAtomGroup()
which = atoms._getIndices()[system]
sel = Selection(ag, which, select, atoms.getACSIndex())
elif isinstance(select, AtomSubset):
sel = select
if isinstance(atoms, AtomGroup):
if sel.getAtomGroup() != atoms:
raise ValueError('select and atoms do not match')
system = np.zeros(len(atoms), bool)
system[sel._getIndices()] = True
else:
if atoms.getAtomGroup() != sel.getAtomGroup():
raise ValueError('select and atoms do not match')
elif not sel in atoms:
raise ValueError('select is not a subset of atoms')
idxset = set(atoms._getIndices())
system = np.array([idx in idxset for idx in sel._getIndices()])
else:
raise TypeError('select must be a string or a Selection instance')
other = np.invert(system)
if model.is3d():
system = np.tile(system, (3, 1)).transpose().flatten()
other = np.tile(other, (3, 1)).transpose().flatten()
ss = matrix[system, :][:, system]
if isinstance(model, PCA):
eda = PCA(model.getTitle() + ' reduced')
eda.setCovariance(ss)
return eda, system
so = matrix[system, :][:, other]
os = matrix[other, :][:, system]
oo = matrix[other, :][:, other]
matrix = ss - np.dot(so, np.dot(linalg.inv(oo), os))
if isinstance(model, GNM):
gnm = GNM(model.getTitle() + ' reduced')
gnm.setKirchhoff(matrix)
return gnm, sel
elif isinstance(model, ANM):
anm = ANM(model.getTitle() + ' reduced')
anm.setHessian(matrix)
return anm, sel
elif isinstance(model, PCA):
eda = PCA(model.getTitle() + ' reduced')
eda.setCovariance(matrix)
return eda, sel
def solveEig(M, n_modes=None, zeros=False, turbo=True, is3d=False):
linalg = importLA()
dof = M.shape[0]
expct_n_zeros = 6 if is3d else 1
if n_modes is None:
eigvals = None
n_modes = dof
else:
if n_modes >= dof:
eigvals = None
n_modes = dof
else:
eigvals = (0, n_modes+expct_n_zeros-1)
def _eigh(M, eigvals=None, turbo=True):
if linalg.__package__.startswith('scipy'):
from scipy.sparse import issparse
if eigvals:
turbo = False
if not issparse(M):
values, vectors = linalg.eigh(M, turbo=turbo, eigvals=eigvals)
else:
try:
from scipy.sparse import linalg as scipy_sparse_la
except ImportError:
raise ImportError('failed to import scipy.sparse.linalg, '
'which is required for sparse matrix '
'decomposition')
if eigvals:
j = eigvals[0]
k = eigvals[-1] + 1
else:
j = 0
k = dof
if k >= dof:
k -= 1
LOGGER.warning('Cannot calculate all eigenvalues for sparse matrices, thus '
'the last eigenvalue is omitted. See scipy.sparse.linalg.eigsh '
'for more information')
values, vectors = scipy_sparse_la.eigsh(M, k=k, which='SA')
values = values[j:k]
vectors = vectors[:, j:k]
else:
if n_modes is not None:
LOGGER.info('Scipy is not found, all modes were calculated.')
else:
n_modes = dof
values, vectors = linalg.eigh(M)
return values, vectors
def _calc_n_zero_modes(M):
from scipy.sparse import issparse
if not issparse(M):
w = linalg.eigvalsh(M)
else:
try:
from scipy.sparse import linalg as scipy_sparse_la
except ImportError:
raise ImportError('failed to import scipy.sparse.linalg, '
'which is required for sparse matrix '
'decomposition')
w, _ = scipy_sparse_la.eigsh(M, k=dof-1, which='SA')
n_zeros = sum(w < ZERO)
return n_zeros
values, vectors = _eigh(M, eigvals, turbo)
n_zeros = sum(values < ZERO)
if n_zeros < n_modes + expct_n_zeros:
if n_zeros < expct_n_zeros:
LOGGER.warning('Fewer than %d (%d) zero eigenvalues were calculated.'%(expct_n_zeros, n_zeros))
elif n_zeros > expct_n_zeros:
LOGGER.warning('More than %d (%d) zero eigenvalues were calculated.'%(expct_n_zeros, n_zeros))
else:
LOGGER.warning('More than %d zero eigenvalues were detected.'%expct_n_zeros)
if not zeros:
if n_zeros > expct_n_zeros:
if n_zeros == n_modes + expct_n_zeros and n_modes != dof:
LOGGER.debug('Determing the number of zero eigenvalues...')
# find the actual number of zero modes
n_zeros = _calc_n_zero_modes(M)
LOGGER.debug('%d zero eigenvalues detected.'%n_zeros)
LOGGER.debug('Solving for additional eigenvalues...')
start = min(n_modes+expct_n_zeros, dof-1); end = min(n_modes+n_zeros-1, dof-1)
values_, vectors_ = _eigh(M, eigvals=(start, end))
values = np.concatenate((values, values_))
vectors = np.hstack((vectors, vectors_))
# final_n_modes may exceed len(eigvals) - no need to fix for the sake of the simplicity of the code
final_n_modes = n_zeros + n_modes
eigvals = values[n_zeros:final_n_modes]
eigvecs = vectors[:, n_zeros:final_n_modes]
vars = 1 / eigvals
else:
eigvals = values[:n_modes]
eigvecs = vectors[:, :n_modes]
vars = div0(1, values)
vars[:n_zeros] = 0.
vars = vars[:n_modes]
return eigvals, eigvecs, vars
def calcPairDeformationDist(model, coords, ind1, ind2, kbt=1.):
"""Returns distribution of the deformations in the distance contributed by each mode
for selected pair of residues *ind1* *ind2* using *model* from a :class:`.ANM`.
Method described in [EB08]_ equation (10) and figure (2).
.. [EB08] Eyal E., Bahar I. Toward a Molecular Understanding of
the Anisotropic Response of Proteins to External Forces:
Insights from Elastic Network Models. *Biophys J* **2008** 94:3424-34355.
:arg model: this is an 3-dimensional :class:`NMA` instance from a :class:`.ANM`
calculations.
:type model: :class:`.ANM`
:arg coords: a coordinate set or an object with :meth:`getCoords` method.
Recommended: ``coords = parsePDB('pdbfile').select('protein and name CA')``.
:type coords: :class:`~numpy.ndarray`.
:arg ind1: first residue number.
:type ind1: int
:arg ind2: secound residue number.
:type ind2: int
"""
try:
resnum_list = coords.getResnums()
resnam_list = coords.getResnames()
coords = (coords._getCoords()
if hasattr(coords, '_getCoords') else coords.getCoords())
except AttributeError:
try:
checkCoords(coords)
except TypeError:
raise TypeError('coords must be a Numpy array or an object '
'with `getCoords` method')
if not isinstance(model, NMA):
raise TypeError('model must be a NMA instance')
elif not model.is3d():
raise TypeError('model must be a 3-dimensional NMA instance')
elif len(model) == 0:
raise ValueError('model must have normal modes calculated')
elif model.getStiffness() is None:
raise ValueError('model must have stiffness matrix calculated')
linalg = importLA()
n_atoms = model.numAtoms()
n_modes = model.numModes()
LOGGER.timeit('_pairdef')
r_ij = np.zeros((n_atoms, n_atoms, 3))
r_ij_norm = np.zeros((n_atoms, n_atoms, 3))
for i in range(n_atoms):
for j in range(i + 1, n_atoms):
r_ij[i][j] = coords[j, :] - coords[i, :]
r_ij[j][i] = r_ij[i][j]
r_ij_norm[i][j] = r_ij[i][j] / linalg.norm(r_ij[i][j])
r_ij_norm[j][i] = r_ij_norm[i][j]
eigvecs = model.getEigvecs()
eigvals = model.getEigvals()
D_pair_k = []
mode_nr = []
ind1 = ind1 - resnum_list[0]
ind2 = ind2 - resnum_list[0]
for m in range(6, n_modes):
U_ij_k = [(eigvecs[ind1*3][m] - eigvecs[ind2*3][m]), (eigvecs[ind1*3+1][m] \
- eigvecs[ind2*3+1][m]), (eigvecs[ind1*3+2][m] - eigvecs[ind2*3+2][m])]
D_ij_k = abs(
sqrt(kbt / eigvals[m]) * (np.vdot(r_ij_norm[ind1][ind2], U_ij_k)))
D_pair_k.append(D_ij_k)
mode_nr.append(m)
LOGGER.report('Deformation was calculated in %.2lfs.', label='_pairdef')
return mode_nr, D_pair_k
def reduceModel(model, atoms, select):
"""Returns reduced NMA model. Reduces a :class:`.NMA` model to a subset of
*atoms* matching *select*. This function behaves differently depending on
the type of the *model* argument. For :class:`.ANM` and :class:`.GNM` or
other :class:`.NMA` models, force constant matrix for system of interest
(specified by the *select*) is derived from the force constant matrix for
the *model* by assuming that for any given displacement of the system of
interest, other atoms move along in such a way as to minimize the potential
energy. This is based on the formulation in [KH00]_. For :class:`.PCA`
models, this function simply takes the sub-covariance matrix for selection.
.. [KH00] Hinsen K, Petrescu A-J, Dellerue S, Bellissent-Funel M-C, Kneller GR.
Harmonicity in slow protein dynamics. *Chem Phys* **2000** 261:25-37.
:arg model: dynamics model
:type model: :class:`.ANM`, :class:`.GNM`, or :class:`.PCA`
:arg atoms: atoms that were used to build the model
:type atoms: :class:`.Atomic`
:arg select: an atom selection or a selection string
:type select: :class:`.Selection`, str
:returns: (:class:`.NMA`, :class:`.Selection`)"""
linalg = importLA()
if not isinstance(model, NMA):
raise TypeError('model must be an NMA instance, not {0}'.format(
type(model)))
if not isinstance(atoms, (AtomGroup, AtomSubset, AtomMap)):
raise TypeError('atoms type is not valid')
if len(atoms) <= 1:
raise TypeError('atoms must contain more than 1 atoms')
if isinstance(model, GNM):
matrix = model._kirchhoff
elif isinstance(model, ANM):
matrix = model._hessian
elif isinstance(model, PCA):
matrix = model._cov
else:
raise TypeError('model does not have a valid type derived from NMA')
if matrix is None:
raise ValueError('model matrix (Hessian/Kirchhoff/Covariance) is not '
'built')
which, select = sliceAtoms(atoms, select)
system = np.zeros(model.numAtoms(), dtype=bool)
system[which] = True
if model.is3d():
system = np.repeat(system, 3)
if isinstance(model, PCA):
ss = matrix[system, :][:, system]
eda = PCA(model.getTitle() + ' reduced')
eda.setCovariance(ss)
return eda, system
else:
matrix = _reduceModel(matrix, system)
if isinstance(model, GNM):
gnm = GNM(model.getTitle() + ' reduced')
gnm.setKirchhoff(matrix)
return gnm, select
elif isinstance(model, ANM):
anm = ANM(model.getTitle() + ' reduced')
anm.setHessian(matrix)
return anm, select
elif isinstance(model, PCA):
eda = PCA(model.getTitle() + ' reduced')
eda.setCovariance(matrix)
return eda, select
from .functions import calcENM
from .modeset import ModeSet
__all__ = [
'calcAdaptiveANM', 'AANM_ONEWAY', 'AANM_ALTERNATING', 'AANM_BOTHWAYS',
'AANM_DEFAULT'
]
AANM_ALTERNATING = 0
AANM_ONEWAY = 1
AANM_BOTHWAYS = 2
AANM_DEFAULT = AANM_ALTERNATING
norm = importLA().norm
def checkInput(a, b, **kwargs):
coordsA = getCoords(a)
if isinstance(a, Atomic):
title = a.getTitle()
atoms = a
else:
title = None
atoms = None
coordsB = getCoords(b)
if title is None:
if isinstance(b, Atomic):
def calcModes(self, n_modes=20, zeros=False, turbo=True):
"""Calculate normal modes. This method uses :func:`scipy.linalg.eigh`
function to diagonalize the Hessian matrix. When Scipy is not found,
:func:`numpy.linalg.eigh` is used.
:arg n_modes: number of non-zero eigenvalues/vectors to calculate.
If ``None`` is given, all modes will be calculated.
:type n_modes: int or None, default is 20
:arg zeros: If ``True``, modes with zero eigenvalues will be kept.
:type zeros: bool, default is ``False``
:arg turbo: Use a memory intensive, but faster way to calculate modes.
:type turbo: bool, default is ``True``
"""
if self._hessian is None:
raise ValueError("Hessian matrix is not built or set")
assert n_modes is None or isinstance(n_modes, int) and n_modes > 0, "n_modes must be a positive integer"
assert isinstance(zeros, bool), "zeros must be a boolean"
assert isinstance(turbo, bool), "turbo must be a boolean"
linalg = importLA()
start = time.time()
shift = 5
if linalg.__package__.startswith("scipy"):
if n_modes is None:
eigvals = None
n_modes = self._dof
else:
if n_modes >= self._dof:
eigvals = None
n_modes = self._dof
else:
eigvals = (0, n_modes + shift)
if eigvals:
turbo = False
if isinstance(self._hessian, np.ndarray):
values, vectors = linalg.eigh(self._hessian, turbo=turbo, eigvals=eigvals)
else:
try:
from scipy.sparse import linalg as scipy_sparse_la
except ImportError:
raise ImportError(
"failed to import scipy.sparse.linalg, " "which is required for sparse matrix " "decomposition"
)
try:
values, vectors = scipy_sparse_la.eigsh(self._hessian, k=n_modes + 6, which="SA")
except:
values, vectors = scipy_sparse_la.eigen_symmetric(self._hessian, k=n_modes + 6, which="SA")
else:
if n_modes is not None:
LOGGER.info("Scipy is not found, all modes are calculated.")
values, vectors = linalg.eigh(self._hessian)
n_zeros = sum(values < ZERO)
if n_zeros < 6:
LOGGER.warning("Less than 6 zero eigenvalues are calculated.")
shift = n_zeros - 1
elif n_zeros > 6:
LOGGER.warning("More than 6 zero eigenvalues are calculated.")
shift = n_zeros - 1
if zeros:
shift = -1
self._eigvals = values[1 + shift :]
self._vars = 1 / self._eigvals
self._trace = self._vars.sum()
self._array = vectors[:, 1 + shift :]
self._n_modes = len(self._eigvals)
LOGGER.debug("{0} modes were calculated in {1:.2f}s.".format(self._n_modes, time.time() - start))
from prody import LOGGER
from .anm import ANM
from .gnm import GNM, ZERO
from .rtb import RTB
from .imanm import imANM
from .exanm import exANM
from .editing import extendModel
from .sampling import sampleModes
from prody.atomic import AtomGroup
from prody.measure import calcTransformation, applyTransformation, calcRMSD
from prody.ensemble import Ensemble
from prody.proteins import writePDB, parsePDB, writePDBStream, parsePDBStream
from prody.utilities import createStringIO, importLA, mad
la = importLA()
norm = la.norm
__all__ = ['ClustENM', 'ClustRTB', 'ClustImANM', 'ClustExANM']
class ClustENM(Ensemble):
'''
ClustENMv2 is the new version of ClustENM(v1) conformation sampling algorithm [KZ16]_.
This ANM-based hybrid algorithm requires PDBFixer and OpenMM for performing energy minimization and MD simulations in implicit/explicit solvent.
It is Python 3.6 compatible and has been only tested on Linux machines.
.. [KZ16] Kurkcuoglu Z., Bahar I., Doruker P., ClustENM: ENM-based sampling of essential conformational space at full atomic resolution. *J Chem* **2016** 12(9):4549-4562.
.. [PE17] Eastman P., Swails J., Chodera J.D., McGibbon R.T., Zhao Y., Beauchamp K.A., Wang L.P., Simmonett A.C., Harrigan M.P., Stern C.D., Wiewiora R.P., Brooks B.R., Pande V.S., OpenMM 7: Rapid Development of High Performance Algorithms for Molecular Dynamics. *PLoS Comput Biol* **2017** 13:e1005659.
# -*- coding: utf-8 -*-
""" This module defines a class for identifying contacts."""
import numpy as np
from prody import LOGGER
from prody.atomic import AtomPointer
from prody.utilities import importLA
from .measure import calcCenter
linalg = importLA()
__all__ = ['Transformation', 'applyTransformation', 'alignCoordsets',
'calcRMSD', 'calcTransformation', 'superpose',
'moveAtoms', 'wrapAtoms',
'printRMSD']
class Transformation(object):
"""A class for storing a transformation matrix."""
__slots__ = ['_matrix']
def __init__(self, *args):
"""Either 4x4 transformation *matrix*, or *rotation* matrix and
*translation* vector must be provided at instantiation."""
nargs = len(args)
if nargs == 1:
def calcModes(self, n_modes=20, zeros=False, turbo=True, hinges=True):
"""Calculate normal modes. This method uses :func:`scipy.linalg.eigh`
function to diagonalize the Kirchhoff matrix. When Scipy is not found,
:func:`numpy.linalg.eigh` is used.
:arg n_modes: number of non-zero eigenvalues/vectors to calculate.
If ``None`` is given, all modes will be calculated.
:type n_modes: int or None, default is 20
:arg zeros: If ``True``, modes with zero eigenvalues will be kept.
:type zeros: bool, default is ``False``
:arg turbo: Use a memory intensive, but faster way to calculate modes.
:type turbo: bool, default is ``True``
:arg hinges: Identify hinge sites after modes are computed.
:type hinges: bool, default is ``True``
"""
if self._kirchhoff is None:
raise ValueError('Kirchhoff matrix is not built or set')
assert n_modes is None or isinstance(n_modes, int) and n_modes > 0, \
'n_modes must be a positive integer'
assert isinstance(zeros, bool), 'zeros must be a boolean'
assert isinstance(turbo, bool), 'turbo must be a boolean'
linalg = importLA()
start = time.time()
shift = 0
if linalg.__package__.startswith('scipy'):
if n_modes is None:
eigvals = None
n_modes = self._dof
else:
if n_modes >= self._dof:
eigvals = None
n_modes = self._dof
else:
eigvals = (0, n_modes + shift)
if eigvals:
turbo = False
if isinstance(self._kirchhoff, np.ndarray):
values, vectors = linalg.eigh(self._kirchhoff, turbo=turbo,
eigvals=eigvals)
else:
try:
from scipy.sparse import linalg as scipy_sparse_la
except ImportError:
raise ImportError('failed to import scipy.sparse.linalg, '
'which is required for sparse matrix '
'decomposition')
try:
values, vectors = (
scipy_sparse_la.eigsh(self._kirchhoff,
k=n_modes + 1, which='SA'))
except:
values, vectors = (
scipy_sparse_la.eigen_symmetric(self._kirchhoff,
k=n_modes + 1,
which='SA'))
else:
if n_modes is not None:
LOGGER.info('Scipy is not found, all modes are calculated.')
values, vectors = linalg.eigh(self._kirchhoff)
n_zeros = sum(values < ZERO)
if n_zeros < 1:
LOGGER.warning('Less than 1 zero eigenvalues are calculated.')
shift = n_zeros - 1
elif n_zeros > 1:
LOGGER.warning('More than 1 zero eigenvalues are calculated.')
shift = n_zeros - 1
if zeros:
shift = -1
self._eigvals = values[1+shift:]
self._vars = 1 / self._eigvals
self._trace = self._vars.sum()
self._array = vectors[:, 1+shift:]
self._n_modes = len(self._eigvals)
if hinges:
self.calcHinges()
LOGGER.debug('{0} modes were calculated in {1:.2f}s.'
.format(self._n_modes, time.time()-start))
def calcModes(self, n_modes=20, zeros=False, turbo=True):
"""Calculate normal modes. This method uses :func:`scipy.linalg.eigh`
function to diagonalize the Hessian matrix. When Scipy is not found,
:func:`numpy.linalg.eigh` is used.
:arg n_modes: number of non-zero eigenvalues/vectors to calculate.
If **None** or ``'all'`` is given, all modes will be calculated.
:type n_modes: int or None, default is 20
:arg zeros: If **True**, modes with zero eigenvalues will be kept.
:type zeros: bool, default is **True**
:arg turbo: Use a memory intensive, but faster way to calculate modes.
:type turbo: bool, default is **True**
"""
if self._hessian is None:
raise ValueError('Hessian matrix is not built or set')
if str(n_modes).lower() == 'all':
n_modes = None
assert n_modes is None or isinstance(n_modes, int) and n_modes > 0, \
'n_modes must be a positive integer'
assert isinstance(zeros, bool), 'zeros must be a boolean'
assert isinstance(turbo, bool), 'turbo must be a boolean'
self._clear()
linalg = importLA()
LOGGER.timeit('_anm_calc_modes')
shift = 5
if linalg.__package__.startswith('scipy'):
if n_modes is None:
eigvals = None
n_modes = self._dof
else:
if n_modes >= self._dof:
eigvals = None
n_modes = self._dof
else:
eigvals = (0, n_modes + shift)
if eigvals:
turbo = False
if isinstance(self._hessian, np.ndarray):
values, vectors = linalg.eigh(self._hessian, turbo=turbo,
eigvals=eigvals)
else:
try:
from scipy.sparse import linalg as scipy_sparse_la
except ImportError:
raise ImportError('failed to import scipy.sparse.linalg, '
'which is required for sparse matrix '
'decomposition')
try:
values, vectors = (
scipy_sparse_la.eigsh(self._hessian, k=n_modes+6,
which='SA'))
except:
values, vectors = (
scipy_sparse_la.eigen_symmetric(self._hessian,
k=n_modes+6,
which='SA'))
else:
if n_modes is not None:
LOGGER.info('Scipy is not found, all modes are calculated.')
values, vectors = np.linalg.eigh(self._hessian)
n_zeros = sum(values < ZERO)
if n_zeros < 6:
LOGGER.warning('Less than 6 zero eigenvalues are calculated.')
shift = n_zeros - 1
elif n_zeros > 6:
LOGGER.warning('More than 6 zero eigenvalues are calculated.')
shift = n_zeros - 1
if zeros:
shift = -1
if n_zeros > n_modes:
self._eigvals = values[1+shift:]
else:
self._eigvals = values[1+shift:]
self._vars = 1 / self._eigvals
self._trace = self._vars.sum()
if shift:
self._array = vectors[:, 1+shift:].copy()
else:
self._array = vectors
self._n_modes = len(self._eigvals)
LOGGER.report('{0} modes were calculated in %.2fs.'
.format(self._n_modes), label='_anm_calc_modes')
def showEmbedding(modes, labels=None, trace=True, headtail=True, cmap='prism'):
"""Visualizes Laplacian embedding of Hi-C data.
:arg modes: modes in which loci are embedded. It can only have 2 or 3 modes for the purpose
of visualization.
:type modes: :class:`ModeSet`
:arg labels: a list of integers indicating the segmentation of the sequence.
:type labels: list
:arg trace: if **True** then the trace of the sequence will be indicated by a grey dashed line.
:type trace: bool
:arg headtail: if **True** then a star and a closed circle will indicate the head and the tail
of the sequence respectively.
:type headtail: bool
:arg cmap: the color map used to render the *labels*.
:type cmap: str
"""
V, _ = _getEigvecs(modes, True)
m, n = V.shape
if labels is not None:
if len(labels) != m:
raise ValueError(
'Modes (%d) and the Hi-C map (%d) should have the same number'
' of atoms. Turn off "masked" if you intended to apply the'
' modes to the full map.' % (m, len(labels)))
if n > 3:
raise ValueError(
'This function can only visualize the embedding of 2 or 3 modes.')
from matplotlib.pyplot import figure, plot, scatter
from mpl_toolkits.mplot3d import Axes3D
if n == 2:
la = importLA()
X, Y = V[:, :2].T
R = np.array(range(len(X)))
R = R / la.norm(R)
X *= R
Y *= R
f = figure()
if trace:
plot(X, Y, ':', color=[0.3, 0.3, 0.3])
if labels is None:
C = 'b'
else:
C = labels
scatter(X, Y, s=30, c=C, cmap=cmap)
if headtail:
plot(X[:1], Y[:1], 'k*', markersize=12)
plot(X[-1:], Y[-1:], 'ko', markersize=12)
elif n == 3:
X, Y, Z = V[:, :3].T
f = figure()
ax = Axes3D(f)
if trace:
ax.plot(X, Y, Z, ':', color=[0.3, 0.3, 0.3])
if labels is None:
C = 'b'
else:
C = labels
ax.scatter(X, Y, Z, s=30, c=C, depthshade=True, cmap=cmap)
if headtail:
ax.plot(X[:1], Y[:1], Z[:1], 'k*', markersize=12)
ax.plot(X[-1:], Y[-1:], Z[-1:], 'ko', markersize=12)
if SETTINGS['auto_show']:
showFigure()
return f
def calcModes(self, n_modes=20, zeros=False, turbo=True, hinges=True):
"""Calculate normal modes. This method uses :func:`scipy.linalg.eigh`
function to diagonalize the Kirchhoff matrix. When Scipy is not found,
:func:`numpy.linalg.eigh` is used.
:arg n_modes: number of non-zero eigenvalues/vectors to calculate.
If **None** or ``'all'`` is given, all modes will be calculated.
:type n_modes: int or None, default is 20
:arg zeros: If **True**, modes with zero eigenvalues will be kept.
:type zeros: bool, default is **True**
:arg turbo: Use a memory intensive, but faster way to calculate modes.
:type turbo: bool, default is **True**
:arg hinges: Identify hinge sites after modes are computed.
:type hinges: bool, default is **True**
"""
if self._kirchhoff is None:
raise ValueError('Kirchhoff matrix is not built or set')
if str(n_modes).lower() == 'all':
n_modes = None
assert n_modes is None or isinstance(n_modes, int) and n_modes > 0, \
'n_modes must be a positive integer'
assert isinstance(zeros, bool), 'zeros must be a boolean'
assert isinstance(turbo, bool), 'turbo must be a boolean'
self._clear()
linalg = importLA()
start = time.time()
shift = 0
if linalg.__package__.startswith('scipy'):
if n_modes is None:
eigvals = None
n_modes = self._dof
else:
if n_modes >= self._dof:
eigvals = None
n_modes = self._dof
else:
eigvals = (0, n_modes + shift)
if eigvals:
turbo = False
if isinstance(self._kirchhoff, np.ndarray):
values, vectors = linalg.eigh(self._kirchhoff, turbo=turbo,
eigvals=eigvals)
else:
try:
from scipy.sparse import linalg as scipy_sparse_la
except ImportError:
raise ImportError('failed to import scipy.sparse.linalg, '
'which is required for sparse matrix '
'decomposition')
try:
values, vectors = (
scipy_sparse_la.eigsh(self._kirchhoff,
k=n_modes + 1, which='SA'))
except:
values, vectors = (
scipy_sparse_la.eigen_symmetric(self._kirchhoff,
k=n_modes + 1,
which='SA'))
else:
if n_modes is not None:
LOGGER.info('Scipy is not found, all modes are calculated.')
values, vectors = linalg.eigh(self._kirchhoff)
n_zeros = sum(values < ZERO)
if n_zeros < 1:
LOGGER.warning('Less than 1 zero eigenvalues are calculated.')
shift = n_zeros - 1
elif n_zeros > 1:
LOGGER.warning('More than 1 zero eigenvalues are calculated.')
shift = n_zeros - 1
if zeros:
shift = -1
self._eigvals = values[1+shift:]
self._vars = 1 / self._eigvals
self._trace = self._vars.sum()
self._array = vectors[:, 1+shift:]
self._n_modes = len(self._eigvals)
if hinges:
self.calcHinges()
LOGGER.debug('{0} modes were calculated in {1:.2f}s.'
.format(self._n_modes, time.time()-start))
def solveEig(M, n_modes=None, zeros=False, turbo=True, is3d=False):
linalg = importLA()
dof = M.shape[0]
expct_n_zeros = 6 if is3d else 1
if n_modes is None:
eigvals = None
n_modes = dof
else:
if n_modes >= dof:
eigvals = None
n_modes = dof
else:
eigvals = (0, n_modes+expct_n_zeros-1)
def _eigh(M, eigvals=None, turbo=True):
if linalg.__package__.startswith('scipy'):
from scipy.sparse import issparse
if eigvals:
turbo = False
if not issparse(M):
values, vectors = linalg.eigh(M, turbo=turbo, eigvals=eigvals)
else:
try:
from scipy.sparse import linalg as scipy_sparse_la
except ImportError:
raise ImportError('failed to import scipy.sparse.linalg, '
'which is required for sparse matrix '
'decomposition')
if eigvals:
j = eigvals[0]
k = eigvals[-1] + 1
else:
j = 0
k = dof
if k >= dof:
k -= 1
LOGGER.warning('Cannot calculate all eigenvalues for sparse matrices, thus '
'the last eigenvalue is omitted. See scipy.sparse.linalg.eigsh '
'for more information')
values, vectors = scipy_sparse_la.eigsh(M, k=k, which='SA')
values = values[j:k]
vectors = vectors[:, j:k]
else:
if n_modes is not None:
LOGGER.info('Scipy is not found, all modes were calculated.')
else:
n_modes = dof
values, vectors = linalg.eigh(M)
return values, vectors
def _calc_n_zero_modes(M):
from scipy.sparse import issparse
if not issparse(M):
w = linalg.eigvalsh(M)
else:
try:
from scipy.sparse import linalg as scipy_sparse_la
except ImportError:
raise ImportError('failed to import scipy.sparse.linalg, '
'which is required for sparse matrix '
'decomposition')
w, _ = scipy_sparse_la.eigsh(M, k=dof-1, which='SA')
n_zeros = sum(w < ZERO)
return n_zeros
values, vectors = _eigh(M, eigvals, turbo)
n_zeros = sum(values < ZERO)
if n_zeros < n_modes + expct_n_zeros:
if n_zeros < expct_n_zeros:
LOGGER.warning('Fewer than %d (%d) zero eigenvalues were calculated.'%(expct_n_zeros, n_zeros))
elif n_zeros > expct_n_zeros:
LOGGER.warning('More than %d (%d) zero eigenvalues were calculated.'%(expct_n_zeros, n_zeros))
else:
LOGGER.warning('More than %d zero eigenvalues were detected.'%expct_n_zeros)
if not zeros:
if n_zeros > expct_n_zeros:
if n_zeros == n_modes + expct_n_zeros and n_modes != dof:
LOGGER.debug('Determing the number of zero eigenvalues...')
# find the actual number of zero modes
n_zeros = _calc_n_zero_modes(M)
LOGGER.debug('%d zero eigenvalues detected.'%n_zeros)
LOGGER.debug('Solving for additional eigenvalues...')
start = min(n_modes+expct_n_zeros, dof-1); end = min(n_modes+n_zeros-1, dof-1)
values_, vectors_ = _eigh(M, eigvals=(start, end))
values = np.concatenate((values, values_))
vectors = np.hstack((vectors, vectors_))
# final_n_modes may exceed len(eigvals) - no need to fix for the sake of the simplicity of the code
final_n_modes = n_zeros + n_modes
eigvals = values[n_zeros:final_n_modes]
eigvecs = vectors[:, n_zeros:final_n_modes]
vars = 1 / eigvals
else:
eigvals = values[:n_modes]
eigvecs = vectors[:, :n_modes]
vars = div0(1, values)
vars[:n_zeros] = 0.
vars = vars[:n_modes]
return eigvals, eigvecs, vars
def showEmbedding(modes, labels=None, trace=True, headtail=True, cmap='prism'):
"""Visualizes Laplacian embedding of Hi-C data.
:arg modes: modes in which loci are embedded. It can only have 2 or 3 modes for the purpose
of visualization.
:type modes: :class:`ModeSet`
:arg labels: a list of integers indicating the segmentation of the sequence.
:type labels: list
:arg trace: if **True** then the trace of the sequence will be indicated by a grey dashed line.
:type trace: bool
:arg headtail: if **True** then a star and a closed circle will indicate the head and the tail
of the sequence respectively.
:type headtail: bool
:arg cmap: the color map used to render the *labels*.
:type cmap: str
"""
V, mask = _getEigvecs(modes, True)
m,n = V.shape
if labels is not None:
if len(labels) != m:
raise ValueError('Modes (%d) and the Hi-C map (%d) should have the same number'
' of atoms. Turn off "masked" if you intended to apply the'
' modes to the full map.'
%(m, len(labels)))
if n > 3:
raise ValueError('This function can only visualize the embedding of 2 or 3 modes.')
from matplotlib.pyplot import figure, plot, scatter
from mpl_toolkits.mplot3d import Axes3D
if n == 2:
la = importLA()
X, Y = V[:,:2].T
R = np.array(range(len(X)))
R = R / la.norm(R)
X *= R; Y *= R
f = figure()
if trace:
plot(X, Y, ':', color=[0.3, 0.3, 0.3])
if labels is None:
C = 'b'
else:
C = labels
scatter(X, Y, s=30, c=C, cmap=cmap)
if headtail:
plot(X[:1], Y[:1], 'k*', markersize=12)
plot(X[-1:], Y[-1:], 'ko', markersize=12)
elif n == 3:
X, Y, Z = V[:,:3].T
f = figure()
ax = Axes3D(f)
if trace:
ax.plot(X, Y, Z, ':', color=[0.3, 0.3, 0.3])
if labels is None:
C = 'b'
else:
C = labels
ax.scatter(X, Y, Z, s=30, c=C, depthshade=True, cmap=cmap)
if headtail:
ax.plot(X[:1], Y[:1], Z[:1], 'k*', markersize=12)
ax.plot(X[-1:], Y[-1:], Z[-1:], 'ko', markersize=12)
if SETTINGS['auto_show']:
showFigure()
return f
def calcPairDeformationDist(model, coords, ind1, ind2, kbt=1.):
"""Returns distribution of the deformations in the distance contributed by each mode
for selected pair of residues *ind1* *ind2* using *model* from a :class:`.ANM`.
Method described in [EB08]_ equation (10) and figure (2).
.. [EB08] Eyal E., Bahar I. Toward a Molecular Understanding of
the Anisotropic Response of Proteins to External Forces:
Insights from Elastic Network Models. *Biophys J* **2008** 94:3424-34355.
:arg model: this is an 3-dimensional NMA instance from a :class:`.ANM
calculations.
:type model: :class:`.ANM`
:arg coords: a coordinate set or an object with ``getCoords`` method.
Recommended: coords = parsePDB('pdbfile').select('protein and name CA').
:type coords: :class:`numpy.ndarray`.
:arg ind1: first residue number.
:type ind1: int
:arg ind2: secound residue number.
:type ind2: int
"""
try:
resnum_list = coords.getResnums()
resnam_list = coords.getResnames()
coords = (coords._getCoords() if hasattr(coords, '_getCoords') else
coords.getCoords())
except AttributeError:
try:
checkCoords(coords)
except TypeError:
raise TypeError('coords must be a Numpy array or an object '
'with `getCoords` method')
if not isinstance(model, NMA):
raise TypeError('model must be a NMA instance')
elif not model.is3d():
raise TypeError('model must be a 3-dimensional NMA instance')
elif len(model) == 0:
raise ValueError('model must have normal modes calculated')
elif model.getStiffness() is None:
raise ValueError('model must have stiffness matrix calculated')
linalg = importLA()
n_atoms = model.numAtoms()
n_modes = model.numModes()
LOGGER.timeit('_pairdef')
r_ij = np.zeros((n_atoms,n_atoms,3))
r_ij_norm = np.zeros((n_atoms,n_atoms,3))
for i in range(n_atoms):
for j in range(i+1,n_atoms):
r_ij[i][j] = coords[j,:] - coords[i,:]
r_ij[j][i] = r_ij[i][j]
r_ij_norm[i][j] = r_ij[i][j]/linalg.norm(r_ij[i][j])
r_ij_norm[j][i] = r_ij_norm[i][j]
eigvecs = model.getEigvecs()
eigvals = model.getEigvals()
D_pair_k = []
mode_nr = []
ind1 = ind1 - resnum_list[0]
ind2 = ind2 - resnum_list[0]
for m in xrange(6,n_modes):
U_ij_k = [(eigvecs[ind1*3][m] - eigvecs[ind2*3][m]), (eigvecs[ind1*3+1][m] \
- eigvecs[ind2*3+1][m]), (eigvecs[ind1*3+2][m] - eigvecs[ind2*3+2][m])]
D_ij_k = abs(np.sqrt(kbt/eigvals[m])*(np.vdot(r_ij_norm[ind1][ind2], U_ij_k)))
D_pair_k.append(D_ij_k)
mode_nr.append(m)
LOGGER.report('Deformation was calculated in %.2lfs.', label='_pairdef')
return mode_nr, D_pair_k
def calcPairDeformationDist(model, coords, ind1, ind2, kbt=1., saveFile=False, filename='out', savePlot=False):
"""Return distribution of the deformations in the distance contributed by each mode
for selected pair of residues *ind1* *ind2* using *model* from a :class:`.ANM`.
Method described in [EB08]_ equation (10) and figure (2).
.. [EB08] Eyal E., Bahar I. Toward a Molecular Understanding of
the Anisotropic Response of Proteins to External Forces:
Insights from Elastic Network Models. *Biophys J* **2008** 94:3424-34355.
:arg model: this is an 3-dimensional NMA instance from a :class:`.ANM
calculations.
:type model: :class:`.ANM`
:arg coords: a coordinate set or an object with ``getCoords`` method.
Recommended: coords = parsePDB('pdbfile').select('protein and name CA').
:type coords: :class:`numpy.ndarray`.
:arg ind1: first residue number.
:type ind1: int
:arg ind2: secound residue number.
:type ind2: int
By default results will not be saved to a *filename* file. To save plot and
data file use ``saveFile=True`` and ``savePlot=True``.
"""
try:
resnum_list = coords.getResnums()
resnam_list = coords.getResnames()
coords = (coords._getCoords() if hasattr(coords, '_getCoords') else
coords.getCoords())
except AttributeError:
try:
checkCoords(coords)
except TypeError:
raise TypeError('coords must be a Numpy array or an object '
'with `getCoords` method')
if not isinstance(model, NMA):
raise TypeError('model must be an NMA instance')
elif not model.is3d():
raise TypeError('model must be a 3-dimensional NMA instance')
elif len(model) == 0:
raise ValueError('model must have normal modes calculated')
elif model.getStiffness() is None:
raise ValueError('model must have stiffness matrix calculated')
linalg = importLA()
n_atoms = model.numAtoms()
n_modes = model.numModes()
LOGGER.timeit('_pairdef')
r_ij = np.zeros((n_atoms,n_atoms,3))
r_ij_norm = np.zeros((n_atoms,n_atoms,3))
for i in range(n_atoms):
for j in range(i+1,n_atoms):
r_ij[i][j] = coords[j,:] - coords[i,:]
r_ij[j][i] = r_ij[i][j]
r_ij_norm[i][j] = r_ij[i][j]/linalg.norm(r_ij[i][j])
r_ij_norm[j][i] = r_ij_norm[i][j]
eigvecs = model.getEigvecs()
eigvals = model.getEigvals()
D_pair_k = []
mode_nr = []
ind1 = ind1 - resnum_list[0]
ind2 = ind2 - resnum_list[0]
for m in xrange(6,n_modes):
U_ij_k = [(eigvecs[ind1*3][m] - eigvecs[ind2*3][m]), (eigvecs[ind1*3+1][m] \
- eigvecs[ind2*3+1][m]), (eigvecs[ind1*3+2][m] - eigvecs[ind2*3+2][m])]
D_ij_k = abs(np.sqrt(kbt/eigvals[m])*(np.vdot(r_ij_norm[ind1][ind2], U_ij_k)))
D_pair_k.append(D_ij_k)
mode_nr.append(m)
LOGGER.report('Deformation was calculated in %.2lfs.', label='_pairdef')
if(saveFile == True):
fout = open(filename+".txt", 'w')
for i in xrange(len(mode_nr)):
fout.write("{} {}\n".format(mode_nr[i], D_pair_k[i]))
fout.close()
LOGGER.info('Data file has been saved.')
if(savePlot == True):
import matplotlib
import matplotlib.pylab as plt
matplotlib.rcParams['font.size'] = '16'
fig = plt.figure(num=None, figsize=(12,8), dpi=100, facecolor='w')
plt.plot(mode_nr, D_pair_k)
plt.xlabel('mode (k)', fontsize = '18')
plt.ylabel('d$^k$' '($\AA$)', fontsize = '18')
plt.savefig(filename+'.png', dpi=100)
LOGGER.info('Plot has been saved.')
return mode_nr, D_pair_k
def calcADPAxes(atoms, **kwargs):
"""Return a 3Nx3 array containing principal axes defining anisotropic
displacement parameter (ADP, or anisotropic temperature factor) ellipsoids.
:arg atoms: a ProDy object for handling atomic data
:type atoms: :class:`~.Atomic`
:kwarg fract: For an atom, if the fraction of anisotropic displacement
explained by its largest axis/eigenvector is less than given value,
all axes for that atom will be set to zero. Values
larger than 0.33 and smaller than 1.0 are accepted.
:type fract: float
:kwarg ratio2: For an atom, if the ratio of the second-largest eigenvalue
to the largest eigenvalue axis less than or equal to the given value,
all principal axes for that atom will be returned.
Values less than 1 and greater than 0 are accepted.
:type ratio2: float
:kwarg ratio3: For an atom, if the ratio of the smallest eigenvalue
to the largest eigenvalue is less than or equal to the given value,
all principal axes for that atom will be returned.
Values less than 1 and greater than 0 are accepted.
:type ratio3: float
:kwarg ratio: Same as *ratio3*.
:type ratio: float
Keyword arguments *fract*, *ratio3*, or *ratio3* can be used to set
principal axes to 0 for atoms showing relatively lower degree of
anisotropy.
3Nx3 axis contains N times 3x3 matrices, one for each given atom. Columns
of these 3x3 matrices are the principal axes which are weighted by
square root of their eigenvalues. The first columns correspond to largest
principal axes.
The direction of the principal axes for an atom is determined based on the
correlation of the axes vector with the principal axes vector of the
previous atom.
.. ipython:: python
from prody import *
protein = parsePDB('1ejg')
calphas = protein.select('calpha')
adp_axes = calcADPAxes( calphas )
adp_axes.shape
These can be written in NMD format as follows:
.. ipython:: python
nma = NMA('ADPs')
nma.setEigens(adp_axes)
nma
writeNMD('adp_axes.nmd', nma, calphas)"""
linalg = importLA()
if not isinstance(atoms, Atomic):
raise TypeError('atoms must be of type Atomic, not {0}'.format(
type(atoms)))
anisous = atoms.getAnisous()
if anisous is None:
raise ValueError('anisotropic temperature factors are not set')
n_atoms = atoms.numAtoms()
axes = zeros((n_atoms * 3, 3))
variances = zeros((n_atoms, 3))
stddevs = zeros((n_atoms, 3))
anisou = anisous[0]
element = zeros((3, 3))
element[0, 0] = anisou[0]
element[1, 1] = anisou[1]
element[2, 2] = anisou[2]
element[0, 1] = element[1, 0] = anisou[3]
element[0, 2] = element[2, 0] = anisou[4]
element[1, 2] = element[2, 1] = anisou[5]
vals, vecs = linalg.eigh(element)
# If negative eigenvalues are found (when ADP matrix is not positive
# definite) set them to 0
vals[vals < 0] = 0
variances[0] = vals
vals = vals**0.5
stddevs[0] = vals
axes[0:3, :] = vals * vecs
for i in range(1, n_atoms):
anisou = anisous[i]
element[0, 0] = anisou[0]
element[1, 1] = anisou[1]
element[2, 2] = anisou[2]
element[0, 1] = element[1, 0] = anisou[3]
element[0, 2] = element[2, 0] = anisou[4]
element[1, 2] = element[2, 1] = anisou[5]
vals, vecs = linalg.eigh(element)
# If negative eigenvalues are found (when ADP matrix is not positive
# definite) set them to 0
vals[vals < 0] = 0
variances[i] = vals
vals = vals**0.5
stddevs[i] = vals
# Make sure the direction that correlates with the previous atom
# is selected
vals = vals * sign((vecs * axes[(i - 1) * 3:(i) * 3, :]).sum(0))
axes[i * 3:(i + 1) * 3, :] = vals * vecs
# Resort the columns before returning array
axes = axes[:, [2, 1, 0]]
torf = None
if 'fract' in kwargs:
fract = float(kwargs['fract'])
assert 0.33 < fract < 1.0, 'fract must be > 0.33 and < 1.0'
variances = variances[:, [2, 1, 0]]
torf = variances[:, 0] / variances.sum(1) > fract
elif 'ratio' in kwargs or 'ratio3' in kwargs or 'ratio2' in kwargs:
if 'ratio2' in kwargs:
ratio = float(kwargs['ratio2'])
assert 0 < ratio < 1.0, 'ratio2 must be > 0 and < 1.0'
dim = 1
else:
ratio = float(kwargs.get('ratio', kwargs.get('ratio3')))
assert 0 < ratio < 1.0, 'ratio or ratio3 must be > 0 and < 1.0'
dim = 2
variances = variances[:, [2, 1, 0]]
torf = variances[:, dim] / variances[:, 0] <= ratio
if torf is not None:
torf = tile(torf.reshape((n_atoms, 1)), (1, 3)).reshape(
(n_atoms * 3, 1))
axes = axes * torf
return axes
def calcADPAxes(atoms, **kwargs):
"""Returns a 3Nx3 array containing principal axes defining anisotropic
displacement parameter (ADP, or anisotropic temperature factor) ellipsoids.
:arg atoms: a ProDy object for handling atomic data
:type atoms: :class:`~.Atomic`
:kwarg fract: For an atom, if the fraction of anisotropic displacement
explained by its largest axis/eigenvector is less than given value,
all axes for that atom will be set to zero. Values
larger than 0.33 and smaller than 1.0 are accepted.
:type fract: float
:kwarg ratio2: For an atom, if the ratio of the second-largest eigenvalue
to the largest eigenvalue axis less than or equal to the given value,
all principal axes for that atom will be returned.
Values less than 1 and greater than 0 are accepted.
:type ratio2: float
:kwarg ratio3: For an atom, if the ratio of the smallest eigenvalue
to the largest eigenvalue is less than or equal to the given value,
all principal axes for that atom will be returned.
Values less than 1 and greater than 0 are accepted.
:type ratio3: float
:kwarg ratio: Same as *ratio3*.
:type ratio: float
Keyword arguments *fract*, *ratio3*, or *ratio3* can be used to set
principal axes to 0 for atoms showing relatively lower degree of
anisotropy.
3Nx3 axis contains N times 3x3 matrices, one for each given atom. Columns
of these 3x3 matrices are the principal axes which are weighted by
square root of their eigenvalues. The first columns correspond to largest
principal axes.
The direction of the principal axes for an atom is determined based on the
correlation of the axes vector with the principal axes vector of the
previous atom.
.. ipython:: python
from prody import *
protein = parsePDB('1ejg')
calphas = protein.select('calpha')
adp_axes = calcADPAxes( calphas )
adp_axes.shape
These can be written in NMD format as follows:
.. ipython:: python
nma = NMA('ADPs')
nma.setEigens(adp_axes)
nma
writeNMD('adp_axes.nmd', nma, calphas)"""
linalg = importLA()
if not isinstance(atoms, Atomic):
raise TypeError('atoms must be of type Atomic, not {0}'
.format(type(atoms)))
anisous = atoms.getAnisous()
if anisous is None:
raise ValueError('anisotropic temperature factors are not set')
n_atoms = atoms.numAtoms()
axes = zeros((n_atoms*3, 3))
variances = zeros((n_atoms, 3))
stddevs = zeros((n_atoms, 3))
anisou = anisous[0]
element = zeros((3, 3))
element[0, 0] = anisou[0]
element[1, 1] = anisou[1]
element[2, 2] = anisou[2]
element[0, 1] = element[1, 0] = anisou[3]
element[0, 2] = element[2, 0] = anisou[4]
element[1, 2] = element[2, 1] = anisou[5]
vals, vecs = linalg.eigh(element)
# If negative eigenvalues are found (when ADP matrix is not positive
# definite) set them to 0
vals[vals < 0] = 0
variances[0] = vals
vals = vals**0.5
stddevs[0] = vals
axes[0:3, :] = vals * vecs
for i in range(1, n_atoms):
anisou = anisous[i]
element[0, 0] = anisou[0]
element[1, 1] = anisou[1]
element[2, 2] = anisou[2]
element[0, 1] = element[1, 0] = anisou[3]
element[0, 2] = element[2, 0] = anisou[4]
element[1, 2] = element[2, 1] = anisou[5]
vals, vecs = linalg.eigh(element)
# If negative eigenvalues are found (when ADP matrix is not positive
# definite) set them to 0
vals[vals < 0] = 0
variances[i] = vals
vals = vals**0.5
stddevs[i] = vals
# Make sure the direction that correlates with the previous atom
# is selected
vals = vals * sign((vecs * axes[(i-1)*3:(i)*3, :]).sum(0))
axes[i*3:(i+1)*3, :] = vals * vecs
# Resort the columns before returning array
axes = axes[:, [2, 1, 0]]
torf = None
if 'fract' in kwargs:
fract = float(kwargs['fract'])
assert 0.33 < fract < 1.0, 'fract must be > 0.33 and < 1.0'
variances = variances[:, [2, 1, 0]]
torf = variances[:, 0] / variances.sum(1) > fract
elif 'ratio' in kwargs or 'ratio3' in kwargs or 'ratio2' in kwargs:
if 'ratio2' in kwargs:
ratio = float(kwargs['ratio2'])
assert 0 < ratio < 1.0, 'ratio2 must be > 0 and < 1.0'
dim = 1
else:
ratio = float(kwargs.get('ratio', kwargs.get('ratio3')))
assert 0 < ratio < 1.0, 'ratio or ratio3 must be > 0 and < 1.0'
dim = 2
variances = variances[:, [2, 1, 0]]
torf = variances[:, dim] / variances[:, 0] <= ratio
if torf is not None:
torf = tile(torf.reshape((n_atoms, 1)), (1, 3)).reshape((n_atoms*3, 1))
axes = axes * torf
return axes
# -*- coding: utf-8 -*-
""" This module defines a class for identifying contacts."""
import numpy as np
from prody import LOGGER
from prody.atomic import AtomPointer
from prody.utilities import importLA
from .measure import calcCenter
linalg = importLA()
__all__ = ['Transformation', 'applyTransformation', 'alignCoordsets',
'calcRMSD', 'calcTransformation', 'superpose',
'moveAtoms', 'wrapAtoms',
'printRMSD']
class Transformation(object):
"""A class for storing a transformation matrix."""
__slots__ = ['_matrix']
def __init__(self, *args):
"""Either 4x4 transformation *matrix*, or *rotation* matrix and
*translation* vector must be provided at instantiation."""
nargs = len(args)
if nargs == 1:
def reduceModel(model, atoms, select):
"""Returns reduced NMA model. Reduces a :class:`.NMA` model to a subset of
*atoms* matching *select*. This function behaves differently depending on
the type of the *model* argument. For :class:`.ANM` and :class:`.GNM` or
other :class:`.NMA` models, force constant matrix for system of interest
(specified by the *select*) is derived from the force constant matrix for
the *model* by assuming that for any given displacement of the system of
interest, other atoms move along in such a way as to minimize the potential
energy. This is based on the formulation in [KH00]_. For :class:`.PCA`
models, this function simply takes the sub-covariance matrix for selection.
.. [KH00] Hinsen K, Petrescu A-J, Dellerue S, Bellissent-Funel M-C, Kneller GR.
Harmonicity in slow protein dynamics. *Chem Phys* **2000** 261:25-37.
:arg model: dynamics model
:type model: :class:`.ANM`, :class:`.GNM`, or :class:`.PCA`
:arg atoms: atoms that were used to build the model
:type atoms: :class:`.Atomic`
:arg select: an atom selection or a selection string
:type select: :class:`.Selection`, str
:returns: (:class:`.NMA`, :class:`.Selection`)"""
linalg = importLA()
if not isinstance(model, NMA):
raise TypeError('model must be an NMA instance, not {0}'
.format(type(model)))
if not isinstance(atoms, (AtomGroup, AtomSubset, AtomMap)):
raise TypeError('atoms type is not valid')
if len(atoms) <= 1:
raise TypeError('atoms must contain more than 1 atoms')
if isinstance(model, GNM):
matrix = model._kirchhoff
elif isinstance(model, ANM):
matrix = model._hessian
elif isinstance(model, PCA):
matrix = model._cov
else:
raise TypeError('model does not have a valid type derived from NMA')
if matrix is None:
raise ValueError('model matrix (Hessian/Kirchhoff/Covariance) is not '
'built')
which, select = sliceAtoms(atoms, select)
system = np.zeros(model.numAtoms(), dtype=bool)
system[which] = True
other = np.invert(system)
if model.is3d():
system = np.repeat(system, 3)
other = np.repeat(other, 3)
ss = matrix[system, :][:, system]
if isinstance(model, PCA):
eda = PCA(model.getTitle() + ' reduced')
eda.setCovariance(ss)
return eda, system
so = matrix[system, :][:, other]
os = matrix[other, :][:, system]
oo = matrix[other, :][:, other]
if other.any():
try:
invoo = linalg.inv(oo)
except:
invoo = linalg.pinv(oo)
matrix = ss - np.dot(so, np.dot(invoo, os))
else:
matrix = ss
if isinstance(model, GNM):
gnm = GNM(model.getTitle() + ' reduced')
gnm.setKirchhoff(matrix)
return gnm, select
elif isinstance(model, ANM):
anm = ANM(model.getTitle() + ' reduced')
anm.setHessian(matrix)
return anm, select
elif isinstance(model, PCA):
eda = PCA(model.getTitle() + ' reduced')
eda.setCovariance(matrix)
return eda, select
