msmbuilder.metrics.Dihedral.__init__

Dihedral.__init__(metric='euclidean', p=2, angles='phi/psi', userfilename='DihedralIndices.dat', V=None, VI=None, indices=None)

Create a distance metric to act on torison angles

Parameters:

metric : {‘braycurtis’, ‘canberra’, ‘chebyshev’, ‘cityblock’,

‘correlation’, ‘cosine’, ‘euclidean’, ‘minkowski’, ‘sqeuclidean’, ‘seuclidean’, ‘mahalanobis’, ‘sqmahalanobis’}

Distance metric to equip the vector space with.

angles : {‘phi’, ‘psi’, ‘chi’, ‘omega’, ‘psi/psi’, etc... OR ‘user’ }

A slash separated list of strings specifying the types of angles to compute per residue. The choices are ‘phi’, ‘psi’, ‘chi’, and ‘omega’, or any combination thereof. If angles = ‘user’, indices are taken from the userfilename

userfilename: string, optional :

filename used for angles=user. Default is ‘DihderalIndices.dat’

p : int, optional

p-norm order, used for metric=’minkowski’

V : ndarray, optional

variances, used for metric=’seuclidean’

VI : ndarray, optional

inverse covariance matrix, used for metric=’mahalanobi’

indices : ndarray, optional

N x 4 numpy array of indices to be considered as dihedral angles. If provided, this overrrides the angles argument. The semantics of the array are that each row, indices[i], is an array of length 4 giving (in order) the indices of 4 atoms that together form a dihedral you want to monitor.

See also

fast_cdist, fast_pdist, scipy.spatial.distance