BrianModel

BrianModel is a library of neuron models and ionic currents for the BRIAN simulator. The purpose of BrianModel is to speed up simulation set-up and reduce code duplication across simulation scripts.

Spiking Neuron Templates

Template neurons are defined by the ionic currents that flow through their membrane. Implemented templates include:

• Hodgkin-Huxley pyramidal neuron (leak, sodium and potassium)
• Hodgkin-Huxley pyramidal neuron with CAN receptors (leak, sodium, potassium, m-current, calcium, CAN)
• Hodgkin-Huxley fast-spiking inhibitory hippocampal (leak, sodium, potassium, m-current)

Ionic Currents

Implemented ionic current libraries include:

• Traub and Miles Hodgkin-Huxley (I_Leak, I_K, I_Na) implementation [1]
• M-Current (I_M) implementation described in [2]
• Calcium current (I_L) implementation by Reuveni et al. [3]
• Calcium pump mechanisms (dCa/dt) implementation by Destexhe et al. [4]
• Calcium current (I_T) implementation by Huguenard et al. [5]
• Calcium-activated non-selective current (I_CAN) implementation by Destexhe et al. [4]
• Wang and Buszáki inhibitory Hodgkin-Huxley (I_Leak, I_K, I_Na) implementation [6]
• Kopell inhibitory Hodgkin-Huxley (I_Leak, I_K, I_Na) implementation [7]

The current library is easily extensible by third-party users due to its hierarchical design. The template neurons and their currents are defined as YAML files, which are conveniently parsed by a Python library which acts as an interface to the BRIAN simulator API's.

Installation

1. Download the repository as it is in your home directory ~/brianmodel

2. Create a file called brianmodel.pth in your python path (the path is located somewhere in local/lib/python2.7/site-packages/, be it global (/usr/) or locally depending on your configuration)

3. Include the library by copying the path to brianmodel in the created .pth file as follows: ~/brianmodel/brianmodel/

Sample Usage

Model Parameter File Structure

Your model neuron is defined as a list of currents and their parameters. You will have to create a YAML parameter file containing all the neuron models used in your simulations. The sample file below (which can be found in includes/) defines two model neurons -- pyramidal and fast-spiking inhibitory -- and their associated currents and parameters.

Typically, a neuron is defined by the area of the cell, the conductance across the cell membrane, and a list of transmembranal ionic currents. This takes the form:

neurons:
    model1:
        area: "1e3 * umetre ** 2"
        conductance: "1 * ufarad ** cm ** -2"

        defined: 
                - class: "IonicCurrentHHTraubLeak"
                  name: "I_leak"
                  g: "1e-5 * siemens * cm ** -2"
                  E: "-70 * mV"
                
                - class: "IonicCurrentHHTraubK"
                  name: "I_K"
                  g: "5 * msiemens * cm ** -2"
                  E: "-100 * mV"
                  vT: "-55 * mV"

Here, "model1" is the identifier of the model neuron, and "defined" contains the list of ionic currents. The neuron of type "model1" contains a leak current and a sodium current. Each individual entry in the current list contains the name of the current class to be instantiated, the name used to identify the current in the BRIAN script, and the parameters of that current equation (the conductance "g", the reversal potential "E", and the Traub constant "vT", in the case of "IonicCurrentHHTraubK").

Existing Currents and their Parameters

This library is shipped with existing current implementations and sample parameter files. These can be found in the includes/ directory. The table below summarises the existing ionic current implementations and their parameters used in the YAML files.

Current	Class Name	Parameters
Traub Leak	IonicCurrentHHTraubLeak	g, E, vT
Traub Potassium	IonicCurrentHHTraubK	g, E, vT
Traub Sodium	IonicCurrentHHTraubNa	g, E, vT
M	IonicCurrentMYamada	g, E, tau
Calcium (L-type)	IonicCurrentCaLReuveni	g, E, tau, caInf, kUnit, kFaraday, depth
Calcium (T-Type)	IonicCurrentCaTHuguenard	g, E, tau, caInf, kUnit, kFaraday, depth
CAN	IonicCurrentCANDestexhe	g, E, beta, cac, temp
Wang Leak	IonicCurrentHHWangLeak	g, E
Wang Sodium	IonicCurrentHHWangNa	g, E
Wang Potassium	IonicCurrentHHWangK	g, E
Monoexponential Synapse	IonicCurrentSynExp	g, E, tau

Existing Neuron Templates

This library is shipped with existing neuron templates. The table below summarises the template neuron and the associated parameter file.

Neuron	Parameter File
Excitatory Hodgkin-Huxley (Traub)	paramsHHTraubPyr.yml
Persistent Firing Excitatory Hodgkin Huxley (Giovannini)	paramsHHPersistentFiring.yml
Inhibitory Hodgking-Huxley (Wang)	paramsHHInhibHippocampus.yml
Inhibitory Hodgking-Huxley (Kopell)	paramsHHKopellInh.yml

Defining and Including Currents

Current entries can either be defined or included from existing YAML files. Below is a model parameter file containing both defined and included currents.

neurons:
    pyramidal:
        area: "29e3 * umetre ** 2"
        conductance: "1 * ufarad * cm ** -2"

        currents:
            included: [paramsSynExpExc.yml, paramsSynExpInh.yml]

            defined:
                - class: "IonicCurrentHHTraubLeak"
                  name: "I_leak"
                  g: "1e-5 * siemens * cm ** -2"
                  E: "-70 * mV"

                - class: "IonicCurrentHHTraubK"
                  name: "I_K"
                  g: "5 * msiemens * cm ** -2"
                  E: "-100 * mV"
                  vT: "-55 * mV"

                - class: "IonicCurrentHHTraubNa"
                  name: "I_Na"
                  g: "50 * msiemens * cm ** -2"
                  E: "50 * mV"
                  vT: "-55 * mV"

    interneuron:
        area: "14e3 * umetre ** 2"
        conductance: "1 * ufarad * cm ** -2"

        currents:
            included: [paramsSynExpExc.yml, paramsSynExpInh.yml]

            defined:
                - class: "IonicCurrentHHWangLeak"
                  name: "I_leak"
                  g: "0.1e-3 * siemens * cm ** -2"
                  E: "-65 * mV"

                - class: "IonicCurrentHHWangK"
                  name: "I_K"
                  g: "9e-3 * siemens * cm ** -2"
                  E: "-90 * mV"

                - class: "IonicCurrentHHWangNa"
                  name: "I_Na"
                  g: "35e-3 * siemens * cm ** -2"
                  E: "55 * mV"

This defines a neuron model called "pyramidal" with its area, conductance and its ionic currents: three defined and two included. The defined currents are the leak, sodium and potassium current of the Traub model. The included currents are mono-exponential synaptic currents which are defined in separate files (paramsSynExpExc.yml, paramsSynExpInh.yml):

- class: "IonicCurrentSynExp"
  name: "I_SynE"
  g: "ge"
  E: "0 * mV"
  tau: "5 * ms"

- class: "IonicCurrentSynExp"
  name: "I_SynI"
  g: "gi"
  E: "-80 * mV"
  tau: "10 * ms"

The path to included currents can be either absolute or relative. Parameter files follow standard YAML syntax.

Simulation Script

1. Import the library in your python BRIAN simulation script:

   import brianmodel as bm

2. Read the neuron model parameters file and create the string-formatted model equations from it:

   # Read parameters from file
   filename = "./params.yml"
   
   # Create the BrianModel object with the given parameter file
   mod = bm.BrianModel(filename)
   mod.readParameterFile()
   
   # Get the generated model string
   modeq = mod.getModelString()

3. This creates a dictionary of string-formatted model equations which you can access by key as standard in Pythong.

4. You can now pass the equations to the BRIAN Simulator. The command below creates a population of 100 neurons defined by the model strings contained in the list identified by "pyramidal"

   # Make BRIAN parse equation string
   eqPyram = Equations(modeq['pyramidal'])
   
   # Create a population of 100 neurons with using the equations
   Pyr = NeuronGroup(100, model=eqPyram, threshold=EmpiricalThreshold(threshold= -20 * mV, refractory=3 * ms), implicit=True, freeze=True)

5. What will follow is your standard BRIAN code.

References

1. Traub and Miles, Neuronal Networks of the Hippocampus, Cambridge, 1991
2. Yamada, W. M., Koch, C., & Adams, P. R. (1989). Multiple Channels and Calcium Dynamics. In C. Koch & I. Segev (Eds.), Methods in neuronal modeling (pp. 97–134). MIT Press.
3. Reuveni I, Friedman A, Amitai Y, Gutnick MJ. Stepwise repolarization from Ca2+ plateaus in neocortical pyramidal cells: evidence for nonhomogeneous distribution of HVA Ca2+ channels in dendrites. Journal of Neuroscience, 1993 Nov, 13(11):4609-21.
4. Destexhe, A., Babloyantz, A., and Sejnowski, T. J. (1993). Ionic mechanisms for intrinsic slow oscillations in thalamic relay neurons. Biophysical journal, 65(4):1538{52.
5. Huguenard, J. R., & McCormick, D. A. (1992). Simulation of the Currents Involved in Rhythmic Oscillations in Thalamic Relay Neurons. Journal of Neurophysiology, 68(4). http://jn.physiology.org/content/68/4/1373
6. Wang X-J, Buzsáki G: Gamma oscillation by synaptic inhibition in a hippocampal interneuronal network model. J Neurosci 1996, 16:6402–13.
7. Kopell NJ, Boergers C, Pervouchine D, Malerba P, Tort A: Gamma and theta rhythms in biophysical models of hippocampal circuits. In Hippocampal Microcircuits A Computational Modeler’s Resource Book. Edited by Cutsuridis V, Graham B, Cobb S, Vida I. New York, NY: Springer New York; 2010:423–457.