navis.persistence_distances

navis.persistence_distances(q, t=None, augment=True, normalize=True, bw=0.2, **persistence_kwargs)

Calculate morphological similarity using persistence diagrams.

This works by:

1. Generate persistence points for each neuron.

2. Create a weighted Gaussian from persistence points and sample 100 evenly spaced points to create a feature vector.

3. Calculate Euclidean distance.

Parameters:

• q/t (NeuronList) – Queries and targets, respectively. If t=None will run queries against queries. Neurons should have the same units, ideally nanometers.

• normalize (bool) – If True, will normalized the vector for each neuron to be within 0-1. Set to False if the total number of linear segments matter.

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

• augment (bool) – Whether to augment the persistence vectors with other neuron properties (number of branch points & leafs and cable length).

• **persistence_kwargs – Keyword arguments are passed to navis.persistence_points().

Returns:

distances

Return type:

pandas.DataFrame

See also

navis.persistence_points()

The function to calculate the persistence points.

navis.persistence_vectors()

Use this to get and inspect the actual vectors used here.