A Python simulator for neuron models specified in the SWC format.

First clone the repository to a local directory, and navigate to the directory. Install the package with the following command,

python setup.py install

The first step is to load a SWC file,

from swcsim import SWCNeuron
neuron = SWCNeuron('neuron.swc')

The SWC file is converted to a Networkx graph, in which each node represents a segment in the original SWC file, and each edge represents the link between two segments. A node has three required attributes:

• morphology: the raw data from the SWC file.
• model: a python class implementing a neural computational model, ex. Hodgkin-Huxley model.
• params: parameters of the computational model. If not specified, the default set of parameter will be used.
• states: states variables of the computational model. states must have at least two fields, I and V.

The second step is to create a simulator by calling create_simulator,

simulator = neuron.create_simulator(
    dt,
    group_1, stimulus_1,
    group_2, stimulus_2,
    ...,
    steps = 1000)

The function create_simulator takes variable pairs of group_i and stimulus_i as positional arguments. group_i is a set or an iterator of identifiers of nodes that receive input from stimulus_i. stimulus_i is a python iterator (ex. a numpy.ndarray or a generator) that will be directly injected into nodes as external current. create_simulator also takes keyword arguments steps or durations that corresponds to number of steps to run.

The function create_simulator creates a generator that updates all nodes and yields its caller SWCNeuron instance at each iteration.

Once the simulator is created by calling create_simulator, running simulation is as simple as looping through the resultant generator,

simulator = neuron.create_simulator(...)
for i,_ in enumerate(simulator):
    # record voltage value for a particular node[x]
    V[i] = neuron.graph.node[x]['states'].V

This repository is maintained by following people:

• Chung-Heng Yeh
• Mehmet Kerem Turkcan
• Tingkai (Thomas) Liu