# -*- coding: utf-8 -*-
"""This module defines a class and a function for anisotropic network model
(ANM) calculations."""
import numpy as np
from prody import LOGGER
from prody.atomic import Atomic, AtomGroup
from prody.proteins import parsePDB
from prody.utilities import importLA, checkCoords
from prody.kdtree import KDTree
from .nma import NMA
from .gnm import GNMBase, ZERO, checkENMParameters
__all__ = ['ANM', 'calcANM']
class ANMBase(NMA):
def __init__(self, name='Unknown'):
super(ANMBase, self).__init__(name)
self._is3d = True
self._cutoff = None
self._gamma = None
self._hessian = None
def _reset(self):
GNMBase._reset(self)
self._hessian = None
self._is3d = True
def getHessian(self):
"""Return a copy of the Hessian matrix."""
if self._hessian is None:
return None
return self._hessian.copy()
def _getHessian(self):
"""Return the Hessian matrix."""
return self._hessian
def setHessian(self, hessian):
"""Set Hessian matrix. A symmetric matrix is expected, i.e. not a
lower- or upper-triangular matrix."""
if not isinstance(hessian, np.ndarray):
raise TypeError('hessian must be a Numpy array')
elif hessian.ndim != 2 or hessian.shape[0] != hessian.shape[1]:
raise ValueError('hessian must be square matrix')
elif hessian.shape[0] % 3:
raise ValueError('hessian.shape must be (3*n_atoms,3*n_atoms)')
elif hessian.dtype != float:
try:
hessian = hessian.astype(float)
except:
raise ValueError('hessian.dtype must be float')
self._reset()
self._hessian = hessian
self._dof = hessian.shape[0]
self._n_atoms = self._dof / 3
def buildHessian(self, coords, cutoff=15., gamma=1., **kwargs):
"""Build Hessian matrix for given coordinate set.
:arg coords: a coordinate set or an object with ``getCoords`` method
:type coords: :class:`numpy.ndarray`
:arg cutoff: cutoff distance (Å) for pairwise interactions,
default is 15.0 Å, minimum is 4.0 Å
:type cutoff: float
:arg gamma: spring constant, default is 1.0
:type gamma: float, :class:`Gamma`
:arg sparse: elect to use sparse matrices, default is **False**. If
Scipy is not found, :class:`ImportError` is raised.
:type sparse: bool
:arg kdtree: elect to use KDTree for building Hessian matrix,
default is **False** since KDTree method is slower
:type kdtree: bool
Instances of :class:`Gamma` classes and custom functions are
accepted as *gamma* argument.
When Scipy is available, user can select to use sparse matrices for
efficient usage of memory at the cost of computation speed."""
try:
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')
cutoff, g, gamma = checkENMParameters(cutoff, gamma)
self._reset()
self._cutoff = cutoff
self._gamma = g
n_atoms = coords.shape[0]
dof = n_atoms * 3
LOGGER.timeit('_anm_hessian')
if kwargs.get('sparse', False):
try:
from scipy import sparse as scipy_sparse
except ImportError:
raise ImportError('failed to import scipy.sparse, which is '
'required for sparse matrix calculations')
kirchhoff = scipy_sparse.lil_matrix((n_atoms, n_atoms))
hessian = scipy_sparse.lil_matrix((dof, dof))
else:
kirchhoff = np.zeros((n_atoms, n_atoms), 'd')
hessian = np.zeros((dof, dof), float)
if kwargs.get('kdtree', False):
LOGGER.info('Using KDTree for building the Hessian.')
kdtree = KDTree(coords)
kdtree.search(cutoff)
for i, j in kdtree.getIndices():
i2j = coords[j] - coords[i]
dist2 = np.dot(i2j, i2j)
g = gamma(dist2, i, j)
super_element = np.outer(i2j, i2j) * (- g / dist2)
res_i3 = i*3
res_i33 = res_i3+3
res_j3 = j*3
res_j33 = res_j3+3
hessian[res_i3:res_i33, res_j3:res_j33] = super_element
hessian[res_j3:res_j33, res_i3:res_i33] = super_element
hessian[res_i3:res_i33, res_i3:res_i33] = \
hessian[res_i3:res_i33, res_i3:res_i33] - super_element
hessian[res_j3:res_j33, res_j3:res_j33] = \
hessian[res_j3:res_j33, res_j3:res_j33] - super_element
kirchhoff[i, j] = -g
kirchhoff[j, i] = -g
kirchhoff[i, i] = kirchhoff[i, i] - g
kirchhoff[j, j] = kirchhoff[j, j] - g
else:
cutoff2 = cutoff * cutoff
for i in range(n_atoms):
res_i3 = i*3
res_i33 = res_i3+3
i_p1 = i+1
i2j_all = coords[i_p1:, :] - coords[i]
for j, dist2 in enumerate((i2j_all ** 2).sum(1)):
if dist2 > cutoff2:
continue
i2j = i2j_all[j]
j += i_p1
g = gamma(dist2, i, j)
res_j3 = j*3
res_j33 = res_j3+3
super_element = np.outer(i2j, i2j) * (- g / dist2)
hessian[res_i3:res_i33, res_j3:res_j33] = super_element
hessian[res_j3:res_j33, res_i3:res_i33] = super_element
hessian[res_i3:res_i33, res_i3:res_i33] = \
hessian[res_i3:res_i33, res_i3:res_i33] - super_element
hessian[res_j3:res_j33, res_j3:res_j33] = \
hessian[res_j3:res_j33, res_j3:res_j33] - super_element
kirchhoff[i, j] = -g
kirchhoff[j, i] = -g
kirchhoff[i, i] = kirchhoff[i, i] - g
kirchhoff[j, j] = kirchhoff[j, j] - g
LOGGER.report('Hessian was built in %.2fs.', label='_anm_hessian')
self._kirchhoff = kirchhoff
self._hessian = hessian
self._n_atoms = n_atoms
self._dof = dof
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')
[docs]class ANM(ANMBase, GNMBase):
"""Class for Anisotropic Network Model (ANM) analysis of proteins
([PD00]_, [ARA01]_).
See a usage example in :ref:`anm`.
.. [PD00] Doruker P, Atilgan AR, Bahar I. Dynamics of proteins predicted by
molecular dynamics simulations and analytical approaches: Application to
a-amylase inhibitor. *Proteins* **2000** 40:512-524.
.. [ARA01] Atilgan AR, Durrell SR, Jernigan RL, Demirel MC, Keskin O,
Bahar I. Anisotropy of fluctuation dynamics of proteins with an
elastic network model. *Biophys. J.* **2001** 80:505-515."""
def __init__(self, name='Unknown'):
super(ANM, self).__init__()
[docs]def calcANM(pdb, selstr='calpha', cutoff=15., gamma=1., n_modes=20,
zeros=False):
"""Return an :class:`ANM` instance and atoms used for the calculations.
By default only alpha carbons are considered, but selection string helps
selecting a subset of it. *pdb* can be :class:`.Atomic` instance."""
if isinstance(pdb, str):
ag = parsePDB(pdb)
title = ag.getTitle()
elif isinstance(pdb, Atomic):
ag = pdb
if isinstance(pdb, AtomGroup):
title = ag.getTitle()
else:
title = ag.getAtomGroup().getTitle()
else:
raise TypeError('pdb must be an atomic class, not {0}'
.format(type(pdb)))
anm = ANM(title)
sel = ag.select(selstr)
anm.buildHessian(sel, cutoff, gamma)
anm.calcModes(n_modes)
return anm, sel