pymaid.cluster_by_synapse_placement

pymaid.cluster_by_synapse_placement(x, sigma=2000, omega=2000, mu_score=True, restrict_cn=None, remote_instance=None)[source]

Cluster neurons based on their synapse placement.

Distances score is calculated by calculating for each synapse of neuron A: (1) the Eucledian distance to the closest synapse in neuron B and (2) comparing the synapse density around synapse A and B. This is type-sensitive: presynapses will only be matched with presynapses, post with post, etc. The formula is described in Schlegel et al., eLife (2017):

f(i_{s},j_{k}) = \exp(\frac{-d^{2}_{sk}}{2\sigma^{2}}) \exp(-\frac{|n(i_{s})-n(j_{k})|}{n(i_{s})+n(j_{k})})

The synapse similarity score for neurons i and j being the average of f(i_{s},j_{k}) over all synapses s of i. Synapse k is the closest synapse of the same sign (pre/post) in neuron j to synapse s. d^{2}_{sk} is the eucledian distance between these distances. Variable \sigma (sigma) determines what distance between s and k is considered “close”. n(i_{s}) and n(j_{k}) are defined as the number of synapses of neuron i/j that are within given radius \omega (omega) of synapse s and j, respectively (same sign only). This esnures that in cases of a strong disparity between n(i_{s}) and n(j_{k}), the synapse similarity will be close to zero even if the distance between s and k is very small.

Parameters

  • x –

    Neurons as single or list of either:

    1. skeleton IDs (int or str)

    2. neuron name (str, exact match)

    3. annotation: e.g. ‘annotation:PN right’

    4. CatmaidNeuron or CatmaidNeuronList object

  • sigma (int, optional) – Distance in nanometer between synapses that is considered to be “close”.

  • omega (int, optional) – Radius in nanometer over which to calculate synapse density.

  • mu_score (bool, optional) – If True, score is calculated as mean between A->B and B->A comparison.

  • restrict_cn (int | list | None, optional) –

    Restrict to given connector types:

    • 0: presynapses

    • 1: postsynapses

    • 2: gap junctions

    • 3: abutting connectors

    If None, will use all connectors. Use either single integer or list. E.g. restrict_cn=[0, 1] to use only pre- and postsynapses.

  • remote_instance (CatmaidInstance, optional) – If not passed, will try using globally defined. Need to provide if neurons are only skids or annotation(s).

Returns

Object that contains distance matrix and methods to plot dendrograms.

Return type

ClustResults