def redim(array, ndim, shape=None):
"""
Add or remove trailing dimensions from an array by reshaping. Useful for turning N-dimensional data into the 2D
shape required for matrix factorization / clustering, and for reversing this transformation. Returns a view of
the input array.
array : numpy array with 2 or more dimensions.
ndim : int, desired number of dimensions when contracting dimensions.
shape : tuple, desired shape when expanding dimensions.
"""
from numpy import prod
result = None
if (ndim > array.ndim) and (shape is None):
raise ValueError(
"Cannot expand dimensions without supplying a shape argument")
if ndim < array.ndim:
new_shape = (*array.shape[:(ndim - 1)], prod(array.shape[(ndim - 1):]))
result = array.reshape(new_shape)
elif ndim > array.ndim:
new_shape = shape
result = array.reshape(new_shape)
elif ndim == array.ndim:
new_shape = array.shape
result = array.reshape(new_shape)
return result
def invert_scale(scaler, X, value):
import numpy as np
new_row = [x for x in X] + [value]
array = np.array(new_row)
array = array.reshape(1, len(array))
inverted = scaler.inverse_transform(array)
return inverted[0, -1]
def pad_array(array):
copied_index = 0
for i in range(len(array)):
if array[i] != 0:
copied_index = i
else:
array[i] = array[copied_index]
arr = array.reshape((len(array), 1))
return arr
def expand_1d_to_2d_array(array, length, axis=0):
"""
General utility routine to 'extend arbitrary dimensional array into a higher dimension
by duplicating the data along a given 'axis' (default is 0) of size 'length'.
Examples::
>>> a = np.array([1, 2, 3, 4])
>>> expand_1d_to_2d_array(a, 4, axis=0)
[[1 2 3 4]
[1 2 3 4]
[1 2 3 4]
[1 2 3 4]]
>>> a = np.array([1, 2, 3, 4])
>>> expand_1d_to_2d_array(a, 4, axis=1)
[[1 1 1 1]
[2 2 2 2]
[3 3 3 3]
[4 4 4 4]]
:param array:
:param length:
:param axis:
:return:
"""
from numpy.ma import MaskedArray
if axis == 0:
new_shape = (length, array.size)
reshaped = array
else:
new_shape = (array.size, length)
reshaped = array.reshape(array.size, 1)
array_2d = np.broadcast_to(reshaped, new_shape, subok=True)
# Broadcast the mask too (this gets lost otherwise)
if isinstance(array, MaskedArray):
array_2d.mask = np.broadcast_to(reshaped.mask, new_shape)
return array_2d
def expand_1d_to_2d_array(array, length, axis=0):
"""
General utility routine to 'extend arbitrary dimensional array into a higher dimension
by duplicating the data along a given 'axis' (default is 0) of size 'length'.
Examples::
>>> a = np.array([1, 2, 3, 4])
>>> expand_1d_to_2d_array(a, 4, axis=0)
[[1 2 3 4]
[1 2 3 4]
[1 2 3 4]
[1 2 3 4]]
>>> a = np.array([1, 2, 3, 4])
>>> expand_1d_to_2d_array(a, 4, axis=1)
[[1 1 1 1]
[2 2 2 2]
[3 3 3 3]
[4 4 4 4]]
:param array:
:param length:
:param axis:
:return:
"""
from numpy.ma import MaskedArray
if axis == 0:
new_shape = (length, array.size)
reshaped = array
else:
new_shape = (array.size, length)
reshaped = array.reshape(array.size, 1)
array_2d = np.broadcast_to(reshaped, new_shape, subok=True)
# Broadcast the mask too (this gets lost otherwise)
if isinstance(array, MaskedArray):
array_2d.mask = np.broadcast_to(reshaped.mask, new_shape)
return array_2d
def invert_scale(scaler, X, yhat):
"""
Since you "scale" object saves data_min and data_max, as well as feature_range,
you just need to pass it to this function, togheter with the object you want to
inverse scale (yhat) too.
:param scaler: Scaler object from scipy, previously calculated in the same dataset being worked here.
:param X: Matrix line used as input to generate this output prediction (yhat).
:param yhat: Predicted output for your network.
:return: Only the "yhat" data rescaled.
"""
# We have input features in X information. We need to add "y" info
# (prediction, not ground truth or reference) before unscaling (because scaler was made with whole dataset, X and y).
# We call it "row" because it's a array (there's no columns). We'll reshape
# it into a matrix line ahead in this function.
new_row = [x for x in X] + [yhat]
array = numpy.array(new_row)
# Array is vertical, but scaler expects a matrix or matrix line, and that's
# why we reshape it.
array = array.reshape(1, len(array))
inverted = scaler.inverse_transform(array)
return inverted[0, -1]
def expand_1d_to_2d_array(array, length, axis=0):
"""
General utility routine to 'extend arbitrary dimensional array into a higher dimension
by duplicating the data along a given 'axis' (default is 0) of size 'length'.
Examples::
>>> a = np.array([1, 2, 3, 4])
>>> expand_1d_to_2d_array(a, 4, axis=0)
[[1 2 3 4]
[1 2 3 4]
[1 2 3 4]
[1 2 3 4]]
>>> a = np.array([1, 2, 3, 4])
>>> expand_1d_to_2d_array(a, 4, axis=1)
[[1 1 1 1]
[2 2 2 2]
[3 3 3 3]
[4 4 4 4]]
:param array:
:param length:
:param axis:
:return:
"""
from numpy.lib.stride_tricks import broadcast_to
if axis == 0:
array_2d = broadcast_to(array, (length, array.size))
# array_2d = np.lib.stride_tricks.as_strided(array_1d, (length, array_1d.size), (0, array_1d.itemsize))
else:
reshaped = array.reshape(array.size, 1)
array_2d = broadcast_to(reshaped, (array.size, length))
# array_2d = np.lib.stride_tricks.as_strided(array_1d, (array_1d.size, length), (array_1d.itemsize, 0))
return array_2d
def ensure_2d_array(array, flags, **kwargs):
array = require(array, requirements = flags, **kwargs)
if len(array.shape) == 1:
array = array.reshape(-1,array.size)
return array
def parsePSF(filename, title=None, ag=None):
"""Return an :class:`.AtomGroup` instance storing data parsed from X-PLOR
format PSF file *filename*. Atom and bond information is parsed from the
file. If *title* is not given, *filename* will be set as the title of the
:class:`.AtomGroup` instance. An :class:`.AtomGroup` instance may be
provided as *ag* argument. When provided, *ag* must have the same number
of atoms in the same order as the file. Data from PSF file will be added
to the *ag*. This may overwrite present data if it overlaps with PSF file
content. Note that this function does not evaluate angles, dihedrals, and
impropers sections."""
if ag is not None:
if not isinstance(ag, AtomGroup):
raise TypeError('ag must be an AtomGroup instance')
psf = openFile(filename, 'rb')
line = psf.readline()
i_line = 1
while line:
line = line.strip()
if line.endswith('!NATOM'):
n_atoms = int(line.split('!')[0])
break
line = psf.readline()
i_line += 1
if title is None:
title = os.path.splitext(os.path.split(filename)[1])[0]
else:
title = str(title)
if ag is None:
ag = AtomGroup(title)
else:
if n_atoms != ag.numAtoms():
raise ValueError('ag and PSF file must have same number of atoms')
serials = zeros(n_atoms, ATOMIC_FIELDS['serial'].dtype)
segnames = zeros(n_atoms, ATOMIC_FIELDS['segment'].dtype)
resnums = zeros(n_atoms, ATOMIC_FIELDS['resnum'].dtype)
resnames = zeros(n_atoms, ATOMIC_FIELDS['resname'].dtype)
atomnames = zeros(n_atoms, ATOMIC_FIELDS['name'].dtype)
atomtypes = zeros(n_atoms, ATOMIC_FIELDS['type'].dtype)
charges = zeros(n_atoms, ATOMIC_FIELDS['charge'].dtype)
masses = zeros(n_atoms, ATOMIC_FIELDS['mass'].dtype)
lines = psf.readlines(71 * (n_atoms + 5))
if len(lines) < n_atoms:
raise IOError('number of lines in PSF is less than the number of '
'atoms')
for i, line in enumerate(lines):
if i == n_atoms:
break
i_line += 1
if len(line) <= 71:
serials[i] = line[:8]
segnames[i] = line[9:13].strip()
resnums[i] = line[14:19]
resnames[i] = line[19:23].strip()
atomnames[i] = line[24:28].strip()
atomtypes[i] = line[29:35].strip()
charges[i] = line[35:44]
masses[i] = line[50:60]
else:
items = line.split()
serials[i] = items[0]
segnames[i] = items[1]
resnums[i] = items[2]
resnames[i] = items[3]
atomnames[i] = items[4]
atomtypes[i] = items[5]
charges[i] = items[6]
masses[i] = items[7]
i = n_atoms
while 1:
line = lines[i].split()
if len(line) >= 2 and line[1] == '!NBOND:':
n_bonds = int(line[0])
break
i += 1
lines = ''.join(lines[i+1:]) + psf.read(n_bonds/4 * 71)
array = fromstring(lines, count=n_bonds*2, dtype=int, sep=' ')
if len(array) != n_bonds*2:
raise IOError('number of bonds expected and parsed do not match')
psf.close()
ag.setSerials(serials)
ag.setSegnames(segnames)
ag.setResnums(resnums)
ag.setResnames(resnames)
ag.setNames(atomnames)
ag.setTypes(atomtypes)
ag.setCharges(charges)
ag.setMasses(masses)
array = add(array, -1, array)
ag.setBonds(array.reshape((n_bonds, 2)))
return ag
def ensure_2d_array(array, flags, **kwargs):
array = require(array, requirements=flags, **kwargs)
if len(array.shape) == 1:
array = array.reshape(-1, array.size)
return array
def parsePSF(filename, title=None, ag=None):
"""Return an :class:`.AtomGroup` instance storing data parsed from X-PLOR
format PSF file *filename*. Atom and bond information is parsed from the
file. If *title* is not given, *filename* will be set as the title of the
:class:`.AtomGroup` instance. An :class:`.AtomGroup` instance may be
provided as *ag* argument. When provided, *ag* must have the same number
of atoms in the same order as the file. Data from PSF file will be added
to the *ag*. This may overwrite present data if it overlaps with PSF file
content. Note that this function does not evaluate angles, dihedrals, and
impropers sections."""
if ag is not None:
if not isinstance(ag, AtomGroup):
raise TypeError('ag must be an AtomGroup instance')
psf = openFile(filename, 'rb')
line = psf.readline()
i_line = 1
while line:
line = line.strip()
if line.endswith('!NATOM'):
n_atoms = int(line.split('!')[0])
break
line = psf.readline()
i_line += 1
if title is None:
title = os.path.splitext(os.path.split(filename)[1])[0]
else:
title = str(title)
if ag is None:
ag = AtomGroup(title)
else:
if n_atoms != ag.numAtoms():
raise ValueError('ag and PSF file must have same number of atoms')
serials = zeros(n_atoms, ATOMIC_FIELDS['serial'].dtype)
segnames = zeros(n_atoms, ATOMIC_FIELDS['segment'].dtype)
resnums = zeros(n_atoms, ATOMIC_FIELDS['resnum'].dtype)
resnames = zeros(n_atoms, ATOMIC_FIELDS['resname'].dtype)
atomnames = zeros(n_atoms, ATOMIC_FIELDS['name'].dtype)
atomtypes = zeros(n_atoms, ATOMIC_FIELDS['type'].dtype)
charges = zeros(n_atoms, ATOMIC_FIELDS['charge'].dtype)
masses = zeros(n_atoms, ATOMIC_FIELDS['mass'].dtype)
lines = psf.readlines(71 * (n_atoms + 5))
if len(lines) < n_atoms:
raise IOError('number of lines in PSF is less than the number of '
'atoms')
for i, line in enumerate(lines):
if i == n_atoms:
break
i_line += 1
if len(line) <= 71:
serials[i] = line[:8]
segnames[i] = line[9:13].strip()
resnums[i] = line[14:19]
resnames[i] = line[19:23].strip()
atomnames[i] = line[24:28].strip()
atomtypes[i] = line[29:35].strip()
charges[i] = line[35:44]
masses[i] = line[50:60]
else:
items = line.split()
serials[i] = items[0]
segnames[i] = items[1]
resnums[i] = items[2]
resnames[i] = items[3]
atomnames[i] = items[4]
atomtypes[i] = items[5]
charges[i] = items[6]
masses[i] = items[7]
i = n_atoms
while 1:
line = lines[i].split()
if len(line) >= 2 and line[1] == '!NBOND:':
n_bonds = int(line[0])
break
i += 1
lines = ''.join(lines[i + 1:]) + psf.read(n_bonds / 4 * 71)
array = fromstring(lines, count=n_bonds * 2, dtype=int, sep=' ')
if len(array) != n_bonds * 2:
raise IOError('number of bonds expected and parsed do not match')
psf.close()
ag.setSerials(serials)
ag.setSegnames(segnames)
ag.setResnums(resnums)
ag.setResnames(resnames)
ag.setNames(atomnames)
ag.setTypes(atomtypes)
ag.setCharges(charges)
ag.setMasses(masses)
array = add(array, -1, array)
ag.setBonds(array.reshape((n_bonds, 2)))
return ag
def invert_scale(scaler, X, yhat): new_row = [x for x in X] + [yhat] array = numpy.array(new_row) array = array.reshape(1, len(array)) inverted = scaler.inverse_transform(array) return inverted[0, -1]
def parsePSF(filename, title=None, ag=None):
"""Returns an :class:`.AtomGroup` instance storing data parsed from X-PLOR
format PSF file *filename*. Atom and bond information is parsed from the
file. If *title* is not given, *filename* will be set as the title of the
:class:`.AtomGroup` instance. An :class:`.AtomGroup` instance may be
provided as *ag* argument. When provided, *ag* must have the same number
of atoms in the same order as the file. Data from PSF file will be added
to the *ag*. This may overwrite present data if it overlaps with PSF file
content. Note that this function does not evaluate angles, dihedrals, and
impropers sections."""
if ag is not None:
if not isinstance(ag, AtomGroup):
raise TypeError("ag must be an AtomGroup instance")
psf = openFile(filename, "rb")
line = psf.readline()
i_line = 1
while line:
line = line.strip()
if line.endswith(b"!NATOM"):
n_atoms = int(line.split(b"!")[0])
break
line = psf.readline()
i_line += 1
if title is None:
title = os.path.splitext(os.path.split(filename)[1])[0]
else:
title = str(title)
if ag is None:
ag = AtomGroup(title)
else:
if n_atoms != ag.numAtoms():
raise ValueError("ag and PSF file must have same number of atoms")
serials = zeros(n_atoms, ATOMIC_FIELDS["serial"].dtype)
segnames = zeros(n_atoms, ATOMIC_FIELDS["segment"].dtype)
resnums = zeros(n_atoms, ATOMIC_FIELDS["resnum"].dtype)
resnames = zeros(n_atoms, ATOMIC_FIELDS["resname"].dtype)
atomnames = zeros(n_atoms, ATOMIC_FIELDS["name"].dtype)
atomtypes = zeros(n_atoms, ATOMIC_FIELDS["type"].dtype)
charges = zeros(n_atoms, ATOMIC_FIELDS["charge"].dtype)
masses = zeros(n_atoms, ATOMIC_FIELDS["mass"].dtype)
# lines = psf.readlines(71 * (n_atoms + 5))
n = 0
n_bonds = 0
for i, line in enumerate(psf):
if line.strip() == b"":
continue
if b"!NBOND:" in line.upper():
items = line.split()
n_bonds = int(items[0])
break
if n + 1 > n_atoms:
continue
if len(line) <= 71:
serials[n] = line[:8]
segnames[n] = line[9:13].strip()
resnums[n] = line[14:19]
resnames[n] = line[19:23].strip()
atomnames[n] = line[24:28].strip()
atomtypes[n] = line[29:35].strip()
charges[n] = line[35:44]
masses[n] = line[50:60]
else:
items = line.split()
serials[n] = items[0]
segnames[n] = items[1]
resnums[n] = items[2]
resnames[n] = items[3]
atomnames[n] = items[4]
atomtypes[n] = items[5]
charges[n] = items[6]
masses[n] = items[7]
n += 1
if n < n_atoms:
raise IOError("number of lines in PSF is less than the number of " "atoms")
# i = n_atoms
# while 1:
# line = lines[i].split()
# if len(line) >= 2 and line[1] == '!NBOND:':
# n_bonds = int(line[0])
# break
# i += 1
# lines = ''.join(lines[i+1:]) + psf.read(n_bonds/4 * 71)
lines = []
for i, line in enumerate(psf):
if line.strip() == b"":
continue
if b"!" in line:
break
lines.append(line.decode(encoding="UTF-8"))
lines = "".join(lines)
array = fromstring(lines, count=n_bonds * 2, dtype=int, sep=" ")
if len(array) != n_bonds * 2:
raise IOError("number of bonds expected and parsed do not match")
psf.close()
ag.setSerials(serials)
ag.setSegnames(segnames)
ag.setResnums(resnums)
ag.setResnames(resnames)
ag.setNames(atomnames)
ag.setTypes(atomtypes)
ag.setCharges(charges)
ag.setMasses(masses)
array = add(array, -1, array)
ag.setBonds(array.reshape((n_bonds, 2)))
return ag
def calcCrossCorr(modes, n_cpu=1, norm=True):
"""Returns cross-correlations matrix. For a 3-d model, cross-correlations
matrix is an NxN matrix, where N is the number of atoms. Each element of
this matrix is the trace of the submatrix corresponding to a pair of atoms.
Cross-correlations matrix may be calculated using all modes or a subset of modes
of an NMA instance. For large systems, calculation of cross-correlations
matrix may be time consuming. Optionally, multiple processors may be
employed to perform calculations by passing ``n_cpu=2`` or more."""
if not isinstance(n_cpu, int):
raise TypeError('n_cpu must be an integer')
elif n_cpu < 1:
raise ValueError('n_cpu must be equal to or greater than 1')
if not isinstance(modes, (Mode, Vector, NMA, ModeSet)):
if isinstance(modes, list):
try:
is3d = modes[0].is3d()
except:
raise TypeError(
'modes must be a list of Mode or Vector instances, '
'not {0}'.format(type(modes)))
else:
raise TypeError(
'modes must be a Mode, Vector, NMA, or ModeSet instance, '
'not {0}'.format(type(modes)))
else:
is3d = modes.is3d()
if is3d:
model = modes
if isinstance(modes, (Mode, ModeSet)):
model = modes._model
if isinstance(modes, (Mode)):
indices = [modes.getIndex()]
n_modes = 1
else:
indices = modes.getIndices()
n_modes = len(modes)
elif isinstance(modes, Vector):
indices = [0]
n_modes = 1
else:
n_modes = len(modes)
indices = np.arange(n_modes)
array = model._getArray()
n_atoms = model._n_atoms
if not isinstance(modes, Vector):
variances = model._vars
else:
array = array.reshape(-1, 1)
variances = np.ones(1)
if n_cpu == 1:
s = (n_modes, n_atoms, 3)
arvar = (array[:, indices] * variances[indices]).T.reshape(s)
array = array[:, indices].T.reshape(s)
covariance = np.tensordot(array.transpose(2, 0, 1),
arvar.transpose(0, 2, 1),
axes=([0, 1], [1, 0]))
else:
import multiprocessing
n_cpu = min(multiprocessing.cpu_count(), n_cpu)
queue = multiprocessing.Queue()
size = n_modes / n_cpu
for i in range(n_cpu):
if n_cpu - i == 1:
indices = modes.indices[i * size:]
else:
indices = modes.indices[i * size:(i + 1) * size]
process = multiprocessing.Process(target=_crossCorrelations,
args=(queue, n_atoms, array,
variances, indices))
process.start()
while queue.qsize() < n_cpu:
time.sleep(0.05)
covariance = queue.get()
while queue.qsize() > 0:
covariance += queue.get()
else:
covariance = calcCovariance(modes)
if norm:
diag = np.power(covariance.diagonal(), 0.5)
D = np.outer(diag, diag)
covariance = div0(covariance, D)
return covariance
