skbio.stats.composition.pairwise_vlr#

skbio.stats.composition.pairwise_vlr(mat, ids=None, ddof=1, robust=False, validate=True)[source]#

Perform pairwise variance log ratio transformation.

Parameters:

matarray_like of shape (n_compositions, n_components): A matrix of proportions.
idsarray_like of str of shape (n_components,): Component names.
ddofint: Degrees of freedom.
robustbool: Whether to mask zeros at the cost of performance.
validatebool: Whether to validate the distance matrix after construction.

Returns:

skbio.DistanceMatrix if validate=True: Distance matrix of variance log ratio values.
skbio.DissimilarityMatrix if validate=False: Dissimilarity matrix of variance log ratio values.

Notes

Pairwise variance log ratio transformation was described in [1] and [2].

References

[1]

V. Lovell D, Pawlowsky-Glahn V, Egozcue JJ, Marguerat S, Bähler J (2015) Proportionality: A Valid Alternative to Correlation for Relative Data. PLoS Comput Biol 11(3): e1004075. https://doi.org/10.1371/journal.pcbi.1004075

[2]

Erb, I., Notredame, C. How should we measure proportionality on relative gene expression data?. Theory Biosci. 135, 21-36 (2016). https://doi.org/10.1007/s12064-015-0220-8

Examples

>>> import numpy as np
>>> from skbio.stats.composition import pairwise_vlr
>>> mat = np.array([np.exp([1, 2, 2]),
...                 np.exp([2, 3, 6]),
...                 np.exp([2, 3, 12])]).T
>>> dism = pairwise_vlr(mat)
>>> dism.redundant_form()
array([[  0.,   3.,  27.],
       [  3.,   0.,  12.],
       [ 27.,  12.,   0.]])