Online HNN Workshop, Spring 2025

If you are new to HNN and biophysical simulations with NEURON, we recommend you do the following to locally install the fastest version of HNN on your Mac computer:

1. Download and install the Anaconda Python distribution here, using the default settings.
2. Open a "Terminal" application and install "Xcode Command-Line Tools" by running the following command: xcode-select --install
3. Close the Terminal window. If the command says "Command line tools are already installed", move to the next step. If the command takes the time to install itself (approximately 10 minutes), then make sure you restart your computer.
4. After restart, open a new Terminal window. The following commands will remove any existing Conda environments named "hnn-core-env", but it will install HNN with Parallelism/MPI support, which is the fastest version of HNN. Copy, paste, and run the following text: conda activate base
conda env remove -y -n hnn-core-env
conda create -y -q -n hnn-core-env python=3.12
conda activate hnn-core-env
mkdir -p $CONDA_PREFIX/etc/conda/activate.d $CONDA_PREFIX/etc/conda/deactivate.d
echo "export OLD_DYLD_FALLBACK_LIBRARY_PATH=\$DYLD_FALLBACK_LIBRARY_PATH" >> $CONDA_PREFIX/etc/conda/activate.d/env_vars.sh
echo "export DYLD_FALLBACK_LIBRARY_PATH=\$DYLD_FALLBACK_LIBRARY_PATH:\${CONDA_PREFIX}/lib" >> $CONDA_PREFIX/etc/conda/activate.d/env_vars.sh
echo "export DYLD_FALLBACK_LIBRARY_PATH=\$OLD_DYLD_FALLBACK_LIBRARY_PATH" >> $CONDA_PREFIX/etc/conda/deactivate.d/env_vars.sh
echo "unset OLD_DYLD_FALLBACK_LIBRARY_PATH" >> $CONDA_PREFIX/etc/conda/deactivate.d/env_vars.sh
export OLD_DYLD_FALLBACK_LIBRARY_PATH=$DYLD_FALLBACK_LIBRARY_PATH
export DYLD_FALLBACK_LIBRARY_PATH=$LD_LIBRARY_PATH:${CONDA_PREFIX}/lib
conda install -y -q "openmpi>5" mpi4py -c conda-forge
pip install "hnn-core[gui,opt,parallel]"

5. Run the following command, which will print a number to the window. Write down this number somewhere, and later you can enter this number in the "Cores" textbox of the HNN GUI for significant speedup! conda activate hnn-core-env
python -c "import psutil ; print(psutil.cpu_count(logical=False)-1)"
6. HNN should now be installed! You should be able to start the GUI by then running the following command in the Terminal window: conda activate hnn-core-env
hnn-gui
7. Troubleshooting: If you run into any errors, you may still be able to install HNN locally, but without Parallelism/MPI speedup, by running the following command in the Terminal. If you still cannot run the GUI after running the following command, we recommend you use the Google CoLab method mentioned above. You can also follow our full Installation Guide here and ask us for help with installation after the workshop. pip install "hnn-core[gui,opt]"

If you are new to HNN and biophysical simulations with NEURON, we recommend you do the following to locally install the fastest version of HNN on your personal Linux computer. If you have installed NEURON previously, we recommend you uninstall any system-wide package installations of it.

1. Download and install the Anaconda Python distribution here, using the default settings.
2. Open a new terminal emulator window. The following commands will remove any existing Conda environments named "hnn-core-env", but it will install HNN with Parallelism/MPI support, which is the fastest version of HNN. Copy, paste, and run the following text: conda activate base
conda env remove -y -n hnn-core-env
conda create -y -q -n hnn-core-env python=3.12
conda activate hnn-core-env
mkdir -p $CONDA_PREFIX/etc/conda/activate.d $CONDA_PREFIX/etc/conda/deactivate.d
echo "export OLD_LD_LIBRARY_PATH=\$LD_LIBRARY_PATH" >> $CONDA_PREFIX/etc/conda/activate.d/env_vars.sh
echo "export LD_LIBRARY_PATH=\$LD_LIBRARY_PATH:\${CONDA_PREFIX}/lib" >> $CONDA_PREFIX/etc/conda/activate.d/env_vars.sh
echo "export LD_LIBRARY_PATH=\$OLD_LD_LIBRARY_PATH" >> $CONDA_PREFIX/etc/conda/deactivate.d/env_vars.sh
echo "unset OLD_LD_LIBRARY_PATH" >> $CONDA_PREFIX/etc/conda/deactivate.d/env_vars.sh
export OLD_LD_LIBRARY_PATH=$LD_LIBRARY_PATH
export LD_LIBRARY_PATH=$LD_LIBRARY_PATH:${CONDA_PREFIX}/lib
conda install -y -q "openmpi>5" mpi4py -c conda-forge
pip install "hnn-core[gui,opt,parallel]"

3. Run the following command, which will print a number to the window. Write down this number somewhere, and later you can enter this number in the "Cores" textbox of the HNN GUI for significant speedup! conda activate hnn-core-env
python -c "import psutil ; print(psutil.cpu_count(logical=False)-1)"
4. HNN should now be installed! You should be able to start the GUI by then running the following command in the terminal window: conda activate hnn-core-env
hnn-gui
5. Troubleshooting: If you run into any errors, you may still be able to install HNN locally, but without Parallelism/MPI speedup, by running the following command in the terminal. If you still cannot run the GUI after running the following command, we recommend you use the Google CoLab method mentioned above. You can also follow our full Installation Guide here and ask us for help with installation after the workshop. pip install "hnn-core[gui,opt]"

If you are using a Windows machine, we strongly recommend you use the Google CoLab method above, since HNN runs slower on Windows, and we currently do not support Parallelism/MPI on Windows. If you still want to install HNN locally on Windows, follow these directions:

1. Download and install the Anaconda Python distribution here, using the default settings.
2. Download and install the Windows-specific NEURON binaries here, using the default settings.
3. Open the application "Anaconda Navigator" and create a new Python environment with Python version 3.12 or below.
4. Open a Terminal in the new Python environment and run the following command: pip install "hnn-core[gui,opt]"
5. HNN should now be installed! You should be able to start the GUI by then running the following command in the terminal window: hnn-gui
6. Troubleshooting: If you run into any errors, then please use the Google CoLab method above.

• Refresh the HNN GUI to restore the default state.
• Click Load data.
• From the "workshops/erp_gui_walkthrough" directory of the "hnn-data" folder, select the experimental_S1_SupraT.txt data file.
• Click Run to run the "default" simulation
• Click the Visualization tab. This should bring you to the tab "ax1" of "Figure 2", corresponding to the lower subplot of Figure 2.
• Under the "ax1" tab, click on the downdown box next to Data to compare and select experimental_S1_SupraT.
• Click the Clear axis button at the bottom of the "ax1" tab
• Click the Add plot button

• From the Visualization tab, adjust the Dipole Smooth Window (ms) in the dipole plot produced by your simulation.
• If you're following from the Exercise 1 Starting Point outlined above, this would correspond to the "ax1" tab of "Figure 2".
• Note that "ax1" in this case refers to the lower subplot of Figure 2 and has a "Type" of "current dipole"
• If you need to plot the data, ensure that Data to compare is set to experimental_S1_SupraT. Otherwise, set it to None.
• Try each of the following values:
  • Value 1:  0
  • Value 2: 15
  • Value 3: 90
• After each update to the value of Dipole Smooth Window, click Add plot to overlay the new dipole on your figure.
• Alternatively, you can first click Clear axis and then Add plot if you do not want to overlay your new dipole on the same plot.

• Reset the Dipole Smooth Window (ms) to its default value of 30
• Click Clear axis
• Ensure the Simulation Data field is set to default, and the Data to Compare field is set to experimental_S1_SupraT
• Next, click Add plot. This should reset your figure to the state it was in before Exercise 1.1.
• Set the Data to Compare field to None
• Next, adjust the Simulation Dipole Scaling in your figure.
• Try each of the following values:
  • Value 1:  1500
  • Value 2:  5000
  • Value 3: 10000
• After each update to the value of Simulation Dipole Scaling, click Add plot to overlay the new dipole on your figure.

Let's now load a different set of evoked drives that was tuned for the suprathreshold experimental condition and compare the new simulation to the default simulation.

• Select the External Drives tab
• Click the Load external drives button.
• From the "workshops/erp_gui_walkthrough" directory of the "hnn-data" folder, select the ERPSupraT_All_Drives.json parameter file.
• In the Simulation tab, change the name of the simulation to All_Drives
• Click Run to start the simulation
• Once your simulation has finished, navigate to the Visualization tab.
• From your newly-generated figure, click on the Data to compare dropdown box and select experimental_S1_SupraT
• Click Clear axis and then Add plot

• If needed, refresh the page to return the HNN GUI to the default state. Note that previous simulations will be cleared when refreshing the page.
• Click Load data
• From the "workshops/erp_gui_walkthrough" directory of the "hnn-data" folder, select the experimental_S1_SupraT.txt data file.
• Click Run to run the "default" simulation.
• From the External drives tab, click the Load external drives button and select the select the ERPSupraT_All_Drives.json parameter file from the "workshops/erp_gui_walkthrough" directory of the "hnn-data" folder.
• In the Simulation tab, change the name of the simulation to All_Drives
• Click Run to run the "All_Drives" simulation.

In this exercise, we'll be adjusting the AMPA weights for the first proximal drive to the Layer 2/3 and Layer 5 pyramidal neurons, and we'll explore how changing these variables affects our simulation.

Since we'll be looking at the first proximal drive only, as noted above, we'll first need to remove the other drives and re-run the simulation.

• From the External Drives tab, delete the drives labelled evdist1 and evprox2.
• You can do so by selecting the drive and then clicking the Delete button at the bottom of the drive parameters.
• Alternatively, you can click the Load external drives button and load the ERPSupraT_Prox1.json parameter file from "hnn-data"
• From the Simulation tab, change the "Name" field to Prox1_Only
• Click the Run button to run the "Prox1_Only" simulation.
• From the Visualization tab, overlay the experimental data on your newly-generated figure by setting Data to compare to experimental_S1_SupraT
• Click Clear axis and then Add plot

One you've added the data to your figure, select the drive labelled evprox1 from the list of drives in the External drives tab. You should now see all of the editable drive parameters for the first proximal drive.

Next, adjust the values of the Layer 2/3 (L2) and Layer 5 (L5) pyramidal cell AMPA weights. Try the following combinations:

• Simulation 01:
  • Name (suggested): Prox1_Only_01
  • L5_pyramidal AMPA weight ("evprox1"): 0.2400
  • L2_pyramidal AMPA weight ("evprox1"): 1.0515
• Simulation 02:
  • Name (suggested): Prox1_Only_02
  • L5_pyramidal AMPA weight ("evprox1"): 0.0480
  • L2_pyramidal AMPA weight ("evprox1"): 0.2103

Once you've run the two simulations above, create a new Figure that displays all of the simulated dipoles and the experimental data on the same graph.

You can do this by selecting any previously-generated figure and adding your new simulations to the dipole plot. For example, if you're working from the Starting Point outlined above:

• Select "ax1" of "Figure 2" from the Visualization tab
• Click the Simulation Data dropdown and select one of your new simulations.
• If you need to plot the data, ensure that Data to compare is set to experimental_S1_SupraT. Otherwise, set it to None.
• Click the Add plot button.
• Repeat these steps for each simulation you would like to compare.

Next, compare and contrast the dipoles and the spiking profiles for the initial simulation (before changing the AMPA weights) and your "best" simulation (after changing the AMPA weights).

You can create a raster plot with the layer-specific dipoles overlaid by following the steps below:

• Select Dipole Layers-Spikes (1x1) from the "Layout template" dropdown menu in the Visualization tab.
• Doing so will trigger the Dataset dropdown box to appear below the "Layout template" dropdown.
• From the "Dataset" dropdown box, select your simulation and then click Make figure.
• Repeat these steps for each simulation you want to visualize

Next, we'll "turn off" the driving inputs to the Layer 2/3 and Layer 5 basket cells by setting their weights to 0, and we'll see how that changes our results. Before running the simulation, take a moment to consider what you think will happen to the dipole and to the spiking.

• From the External Drives tab, first set the AMPA weights for the Layer 2/3 and Layer 5 pyramidal neurons to the values from Exercise 2.1 that produced the "best" fit to the "M25" peak.
• Next, set the AMPA and NMDA weights to 0 for the L5_basket and L2_basket cells in the evprox1 drive parameters.
• From the Simulation tab, change the value of the "Name" field to Prox1_Only_noBasket
• Click Run to run the "Prox1_Only_noBasket" simulation.

You can create a raster plot with the layer-specific dipoles overlaid by following the steps in Exercise 2.2 above.

• Follow all instructions in the Exercise 2 Starting Point section above
• From the External drives tab, click the Load external drives button and select the select the ERPSupraT_Prox1.json parameter file from the "workshops/erp_gui_walkthrough" directory of the "hnn-data" folder.
• From the Simulation tab, change the "Name" field to Prox1_Only
• Click the Run button to run the "Prox1_Only" simulation.

In this exercise, we'll be adjusting the timing of the first distal drive, and we'll explore how doing so affects our simulation.

This time, we'll start by load in a parameter set that includes the "tuned" version of the first proximal drive, as well as the un-tuned distal drive.

• Select the External Drives tab.
• Click the Load external drives button. From the "workshops/erp_gui_walkthrough" directory of the "hnn-data" folder, select the SupraT_Prox1_Dist1.json parameter file.
• From the Simulation tab, change the value of the "Name" field to P1_D1
• Click Run to run the "P1_D1" simulation.
• From the Visualization tab, overlay the experimental data on your newly-generated figure by setting Data to compare to experimental_S1_SupraT
• Click Clear axis and then Add plot

Next, we'll try adjusting the standard deviation of the distal drive. Note that this is only one parameter that affects the timing of the driving input.

Before running the simulations below, take a moment to consider what you think will happen to the dipole and to the spiking.

• From the External Drives tab, select evdist1. You should now see all of the editable drive parameters for the distal drive.
• Adjust the value of the standard deviation around the mean arrival time, listed as: Std dev time.
• Try the following combinations:
• Simulation 01:
  • Name (suggested): P1_D1_01
  • Std dev time: 13.81
• Simulation 02:
  • Name (suggested): P1_D1_02
  • Std dev time: 3.81

Compare the layer-specific dipoles and the spiking profiles for each simulation, as done in Exercise 2.2 above.

• Follow all instructions in the Exercise 2 Starting Point and Exercise 3 Starting Point sections above
• From the External drives tab, click the Load external drives button and select the select the ERPSupraT_Prox1_Dist1.json parameter file from the "workshops/erp_gui_walkthrough" directory of the "hnn-data" folder.
• From the Simulation tab, change the "Name" field to P1_D1
• Click the Run button to run the "P1_D1" simulation.

In this exercise, we'll be adjusting the timing of the second proximal drive, and we'll explore how doing so affects our simulation.

This time, we'll start by load in a parameter set that includes the "tuned" version of the first proximal and first distal drives, as well as the "un-tuned" second proximal drive.

• Select the External Drives tab.
• Click the Load external drives button. From the "workshops/erp_gui_walkthrough" directory of the "hnn-data" folder, select the SupraT_Prox1_Dist1_Prox2.json parameter file.
• From the Simulation tab, change the value of the "Name" field to P1D1P2
• Click Run to run the "P1D1P2" simulation.
• From the Visualization tab, overlay the experimental data on your newly-generated figure by setting Data to compare to experimental_S1_SupraT
• Click Clear axis and then Add plot

Next, we'll try adjusting the mean arrival time of the second proximal drive, which is another parameter (in addition to the standard deviation) that affects the timing of the driving inputs.

Before running the simulations below, take a moment to consider what you think will happen to the dipole and to the spiking.

• From the External Drives tab, select evprox2. You should now see all of the editable drive parameters for the second proximal drive.
• Adjust the value of the mean arrival time of the drive, listed as: Mean time.
• Try the following combinations:
• Simulation 01:
  • Name (suggested): P1D1P2_01
  • Mean time: 105
• Simulation 02:
  • Name (suggested): P1D1P2_02
  • Std dev time: 95

Compare the layer-specific dipoles and the spiking profiles for each simulation, as done in Exercise 2.2 above.