Stimulation of a Neuron-Astrocyte Network (SNAN) with Working Memory (WM)

Some parts of the code are from: Working memory with self-repairing

This repository contains MATLAB code for online/offline excitatory/inhibitory stimulation of a Neuron-Astrocyte Network (SNAN) with Working Memory (WM), published in iScience journal.

Link to the paper: ...

Requirements

• The code was tested on MATLAB 2022b, it might work with earlier versions.
• The minimum required amount of RAM is 32 GB.

Graphical Abstract

Video performance of the model

Click the image to watch the video

Settings

Model's configuration

Quick setup

There are 3 experiments in the paper, each requiring a unique set of settings. To simplify this for users, we have grouped the variables that need to be changed in a switch statement inside model_parameters.m. All you need to do is change the params.simPattern variable at the top of the model_parameters.m file.

params.simPattern = 1; % Experiment 1 - Enhancement (default: damaged network)
params.simPattern = 2; % Experiment 2 - Disruption (default: disrupts both train and test)
params.simPattern = 3; % Experiment 3 - Augmentation

• To start the simulation, run the main.m file. When you run this file, you will get a GUI pop-up to confirm your settings before start of the simulation. You have to press continue to start the simulation otherwise the simulation will not start.
• To plot the outputs, open the plot_settings.m file, set the variables you want to plot to 1, and run the file.
• For video, when you get the result, in the video window, go to "playback" -> "frame rate" and increase the frame rate to 500 or 1000 and check the "allow frame drop ..." box. This increases the speed of video playback.
• In the plot_settings, make sure that the variable you are trying to plot is with the dimension of params.n in the init_model.m.

Experiment 1 - Enhancement

To get the results for this experiment, set the following parameters in the model_parameters.m file:

params.simPattern = 3;
params.impairmode = 1; % For healthy network.
params.impairmode = 3; % For damaged network (default).

Configuration Details:

Setting simPattern to 3 activates the Enhancement mode (Case 3). This configuration:

• Sets the simulation duration (t_end) to 5.6s.
• Enables Synaptic Damage Mode 3 (impairmode = 3), which applies concentrated damage based on a specific damaging pattern.
• Sets the learning and testing impulse order to [1,2,3,4].
• Enables Automatic Stimulation (stimulation_mode = 1) targeting Excitatory neurons (stimulation_neuro_E = 1) at 3.2s to counteract any damage. Activating these two parameters uses the stimulator_enhance function in simulate_model.m.

Since this experiment involves two parts, a healthy network and a damaged network, you can enable synaptic damage by setting impairmode = 3, which applies concentrated damage based on a specific damaging pattern. For a healthy network, set impairmode = 0. In healthy mode, even though automatic stimulation is active, the Enhancement algorithm recognizes this and does not activate any stimulating electrodes.

Experiment 2 - Disruption

To get the results for this experiment, set the following parameters in the model_parameters.m file:

params.simPattern = 4;
params.disruptBothTrainTest = 0; % Disrupts only in the training phase.
params.disruptBothTrainTest = 1; % Disrupts both in train and test phases (default).

Configuration Details: Setting simPattern to 4 activates the Disruption mode (Case 4). This configuration:

• Sets the simulation duration (t_end) to 3.5s.
• Sets the learning and testing impulse order to [9, 3, 11, 7].
• Maintains a Healthy Network (impairmode = 0) but applies external disruption.
• Sets disruption pattern indices to [3,7]. To disrupt other patterns, change the disruption_pattern variable.
• Enables Automatic Stimulation (stimulation_mode = 1) targeting Inhibitory neurons (stimulation_neuro_I = 1) at 1.28s to disrupt the pattern recall. Activating these two parameters uses the stimulator_dist function in simulate_model.m.

Experiment 3 - Augmentation

To get the results for this experiment, set the following parameters in the model_parameters.m file:

params.simPattern = 5;

Configuration Details: Setting simPattern to 5 activates the Augmentation mode (Case 5). This configuration:

• Sets the simulation duration (t_end) to 2.6s.
• Sets learning order to [4, 5] and testing order to [4, 2, 5, 1].
• Maintains a Healthy Network (impairmode = 0).
• Switches to Manual Stimulation (stimulation_mode = 0) with Augmentation enabled (stimulation_Augmentation = 1).
• Activates specific electrodes [15,8,12,16] for stimulation.
• Uses two sets of timing for manual stimulation (stimulation_time_neuro_E_Augmentation = [0.12, 0.42]).
• Activates specific electrodes for each timing [6,10,14] and [3,4,11,12].
• Unlike other modes, there is no separate stimulator function for this mode. Instead, the setup is configured inside the init_model.m file.

---

Code

• The term train in the code is mentioned as sample phase in the paper.
• In all modes, stimulation current (I_stim) is added to I_app_astro inside the get_neuron_astrozone_activity.m function.
• Files with the TB extension are Test Bench files used for fast iteration and testing of a function. They are typically designed for standalone execution but sometimes they need data from a simulation. To get the TB files for enhance and disrupt functions, email the corresponding author (Mahmood Amiri).
• accuracy_elec measures the correlation, mentioned in the paper.
• Extra code for synaptic plasticity is provided in step_neurons.m for future work.
• To load your saved data, open saveLoadData.m and run the Import section.
• To use sendTelegramMessage function you need to set some configuration inside its function.
• To understand the relationship between code blocks, we recommend to check out the AI-generated overview by DeepWiki, linked at the bottom of this readme file. Note that this is a fully automated AI generated document.

Authors

• Pedram Naghieh - Software, Formal analysis, Writing – original draft, Methodology - PedRaMNG
• Mahmood Amiri - Conceptualization, Methodology, Supervision, Writing review & editing
• Herbert Peremans - Conceptualization, Writing review & editing

Cite

P. Naghieh, M. Amiri, and H. Peremans, “Targeted Neuronal Stimulation Modulates Working Memory via Astrocytic Interactions: Insights from a Spiking Neuron-Astrocyte Model”, iScience, p. ... , jan. 2025, doi: ...

License

This project is licensed under the MIT License - see the LICENSE.md file for details.

DeepWiki

AI-generated overview of the code and the relationship between the code blocks: DeepWiki - WM-with-Stimulation