connectome_manipulator.connectome_manipulation.manipulation.syn_prop_alteration.SynapsePropertAlteration

Connectome manipulation class for altering synapse property values:

Modifies the values of a selected synapse property and a selected set of synapses, as given by constant, multiplicative, additive, or random values drawn from chosen distributions. The manipulation can be applied through the apply() method.

connectome_manipulator.connectome_manipulation.manipulation.syn_prop_alteration.SynapsePropertAlteration.apply(self, split_ids, syn_prop, new_value, sel_src=None, sel_dest=None, syn_filter=None, amount_pct=100.0, **kwargs)

Applies a modification of synapse property values on a selected set of synapses.

Parameters:

  • split_ids (list-like) – List of neuron IDs that are part of the current data split; will be automatically provided by the manipulator framework

  • syn_prop (str) – Selected synapse property values of which should be altered; can be selected from all available properties in the given SONATA input edges table (except for “@source_node” and “@target_node”)

  • new_value (dict) – Dictionary specifying the type of alteration; must contain a “mode” key with one of the available alteration modes, together with additional settings depending on the mode, see Notes; optionally, can also contain a “range” key with lower/upper value bounds, an “rng” key to specify a random number generator, and a “kwargs” dict with additional arguments for the random number generator

  • sel_src (str/list-like/dict) – Source (pre-synaptic) neuron selection

  • sel_dest (str/list-like/dict) – Target (post-synaptic) neuron selection

  • syn_filter (dict) – Optional filter dictionary with property names (keys) with a corresponding list of values (e.g., {"layer": [2, 3, 4], "synapse_class": ["EXC"]}); only synapses corresponding to that selection of values will be considered for alteration

  • amount_pct (float) – Percentage of randomly sampled synapses to be altered

  • **kwargs – Additional keyword arguments - Not used

Note

Available alteration modes and corresponding settings:

  • “setval”: Absolute value, as given by “value”

  • “scale”: Multiplicative scaling, as given by “factor”

  • “offset”: Additive offset, as given by “value”

  • “shuffle”: Shuffling of values among synapses; only supported when using a single data split

  • “randval”: Random absolute value drawn from a given distribution (*)

  • “randscale”: Multiplicative scaling by a random value drawn from a given distribution (*)

  • “randadd”: Additive offset by a random value drawn from a given distribution (*)

(*) A random distribution can be specified in new_value by choosing a random number generator as “rng” from numpy.random (e.g., “normal”), together with “kwargs” which will be passed when drawing values from that generator (e.g., “loc” and “scale” in case of “normal”)

Note

Input/output edges (synapse) tables are accessed through the writer object:

  • Loading input edges: edges_table = self.writer.to_pandas()

  • Writing output edges: self.writer.from_pandas(edges_table_manip)