navis.persistence_vectors

navis.persistence_vectors(pers, threshold=None, samples=100, bw=0.2, center=False)

Produce vectors from persistence points.

Works by creating a Gaussian and sampling samples evenly spaced points across it.

Parameters:

• pers (pd.DataFrame | list thereof) – The persistence points (see navis.persistence_points()). For vectors for multiple neurons, provide either a list of persistence points DataFrames or a single DataFrame with a “neuron_id” column.

• threshold (float, optional) – If provided, segments shorter (death - birth) than this will not be used to create the Gaussian.

• samples (int) – Number of points sampled across the Gaussian.

• bw (float) – Bandwidth for Gaussian kernel: larger = smoother, smaller = more detailed.

• center (bool) – Whether to center the individual curves on their highest value. This is done by “rolling” the axis (using np.roll) which means that elements that roll beyond the last position are re-introduced at the first.

Returns:

• vectors (np.ndarray)

• samples (np.ndarray) – Sampled distances. If center=True the absolute values don’t make much sense anymore.

References

Li Y, Wang D, Ascoli GA, Mitra P, Wang Y (2017) Metrics for comparing neuronal tree shapes based on persistent homology. PLOS ONE 12(8): e0182184. https://doi.org/10.1371/journal.pone.0182184

See also

navis.persistence_points()

The function to calculate the persistence points.

navis.persistence_distances()

Get distances based on (augmented) persistence vectors.