Tutorial 4: Network with Multiple Populations

Overview

This tutorial demonstrates simulation of multiple populations consisting of different neuron types.

See also

Corresponding BMTK tutorial: Tutorial: Multi-Population Recurrent Network

Changes from BMTK to bionetlite

Only the import statement needs to be changed. All other code is completely identical.

Before (BMTK):

from bmtk.builder.networks import NetworkBuilder

After (bionetlite):

from bionetlite import NeuliteBuilder as NetworkBuilder

Generated Files

population.csv

Contains information for 100 neurons (80 excitatory + 20 inhibitory).

connection.csv

Contains information for 4 types of synaptic connections (E→E, E→I, I→E, I→I).

Files Generated by Neulite

bionetlite generates files in the following directory:

Directory Structure

sim_ch04_nl/
├── V1_population.csv
├── V1_V1_connection.csv
├── kernel/
│   └── config.h
└── data/
    ├── Scnn1a_473845048_m.swc
    ├── Pvalb_470522102_m.swc
    ├── 472363762_fit.csv
    └── 472912177_fit.csv

V1_population.csv

A population file for Neulite with multiple neuron types defined:

#n_cell,n_comp,name,swc_file,ion_file
80,3682,Scnn1a_100,data/Scnn1a_473845048_m.swc,data/472363762_fit.csv
20,1900,PV_101,data/Pvalb_470522102_m.swc,data/472912177_fit.csv

Each row contains the following information:

  • #n_cell: Number of cells (80 excitatory, 20 inhibitory)

  • n_comp: Number of compartments (Scnn1a: 3682, PV: 1900)

  • name: Population name

  • swc_file: Path to SWC morphology file

  • ion_file: Path to ion channel parameter file

V1_V1_connection.csv

A connection file for Neulite that defines all synaptic connections within and between populations:

#pre nid,post nid,post cid,weight,tau_decay,tau_rise,erev,delay,e/i
0,1,5,5e-05,1.0,3.0,0.0,2,e
0,1,3351,5e-05,1.0,3.0,0.0,2,e
0,2,28,5e-05,1.0,3.0,0.0,2,e
...

Meaning of each column:

  • #pre nid: Presynaptic neuron ID

  • post nid: Postsynaptic neuron ID

  • post cid: Postsynaptic compartment ID

  • weight: Synaptic weight

  • tau_decay: Decay time constant

  • tau_rise: Rise time constant

  • erev: Reversal potential

  • delay: Synaptic delay (ms)

  • e/i: Excitatory (e) or inhibitory (i)

Building and Running the Neulite Kernel

Step 1: Deploy Kernel Sources

Copy the contents of Neulite’s kernel directory to the kernel directory generated by bionetlite. Exclude config.h as it has already been generated by bionetlite:

rsync -av --exclude='config.h' /path/to/neulite/kernel/ sim_ch04_nl/kernel/

Step 2: Build the Kernel

Navigate to the kernel directory and build. config.h has already been generated by bionetlite:

cd sim_ch04_nl/kernel
make

Step 3: Place the Binary

Copy the built binary file nl to the parent directory:

cp nl ../

Step 4: Run the Simulation

Navigate to the directory generated by bionetlite and run the simulation:

cd ..
./nl V1_population.csv V1_V1_connection.csv

Note

About External File Input

Neulite currently does not support spike input from external files as performed in BMTK tutorials. Therefore, simulations using the files generated in this tutorial will not produce spikes due to the absence of external input.

To verify network behavior, please configure constant current injection (I_AMP, I_DELAY, I_DURATION) in config.h.

Simulation Result Files

The following text files are generated after running the Neulite kernel:

s.dat - Spike Time Data

Space-delimited text format recording one spike event per line:

Time(ms) NeuronID
12.5 0
15.3 15
18.9 42
23.1 0
...

  • Column 1: Time at which spike occurred (ms)

  • Column 2: ID of the neuron that generated the spike (0-79: excitatory, 80-99: inhibitory)

v.dat - Membrane Potential Data

Space-delimited text format recording membrane potential of all neurons per line:

Time(ms) Neuron0 Neuron1 ... Neuron99
0.0 -65.0 -65.0 ... -65.0
0.1 -64.8 -64.9 ... -65.1
0.2 -64.6 -64.8 ... -65.0
...

  • Column 1: Time (ms)

  • Column 2 onwards: Soma membrane potential of each neuron (mV), for all 100 neurons

Note

Data for excitatory (0-79) and inhibitory (80-99) neurons are contained in the same file.

About Mixing with LIF Models

Note

Tutorial 4 originally included LIF model neurons as well, but bionetlite and Neulite currently support only biophysical neurons.

bionetlite supports the following neuron types:

  • ✅ biophysical neurons

  • ❌ point neurons (LIF, etc.)

  • ❌ Virtual neurons

Therefore, only files related to biophysical neurons are generated in this tutorial.

Next Steps