This directory contains scripts for simulating the GGN model and the mushroom body olfactory circuit around it (originally in NEURON 7.4 with Python 2.7) related to the article:

"Feedback inhibition and its control in an insect olfactory circuit". Subhasis Ray, Zane Aldworth, and Mark Stopfer; 2020. eLife.

• April 2024: Updated to run with NEURON 8.2.4 on Python 3.12 during COMBINE/Harmony-2024 Hackathon.

Required packages:

• networkx (graph data structures)
• neuron (simulation)
• sklearn (spatial clustering based connections)
• pint (units)
• pyyaml (configuration)
• nsdf (saving data): pip install git+https://github.com/nsdf/nsdf.git
• vtk (for 3D visualization)
• numpy (numerics)
• matplotlib (visualization)

For testing, run simulation scripts from the root directory of this repo (where this README.md file resides). Edit the nrn/nrninit.sh (or nrninit.bat on Windows) with the path for this directory.

If you are running from another directory, edit nrn/nrninit.sh and set MODEL_DIR to the path of that directory.

• To initialize the environment variables run the nrn/nrninit.sh file. For Linux/OSX with bash shell: source nrn/nrninit.sh

• You also have to build the NEURON mechanisms (.mod) files:

  nrnivmodl mb/mod

• To simulate GGN in isolation to study signal attenuation with distance, run:

  python mb/test_cell/ggn_voltage_attenuation_vclamp.py

• To simulate the PN->KC<->GGN network model, create a directory called data for dumping simulated data and run:

  python mb/network/pn_kc_ggn_network.py

  You may want to edit the configuration file mb/network/config.yaml to reduce the number of cells in the network and the duration of the simulation.

• various example scripts for analyzing simulated data.

• nhpp.py : Generate times for nonhomogeneous Poisson process

• GGN_20170309_sc.swc : morphology trace of GGN shrinkage corrected for methylsalicylate fixing
• GGN_20170309_sc.hoc : cell template for GGN model
• kc_1_comp.hoc : single compartmental KC model

• change_pn_spikes.py : take a simulated data file and create another one after modifying the PN spike times, so that same model is simulated with modified spike input.
• config.yaml : configuration file for setting network model parameters
• fixed_network_changing_stim.py : simulate a given network model with different PN spike inputs.
• kc_ggn_feedback_dclamp.py : test of single KC with GGN inhibition when the GGN is driven by a dynamic clamp.
• kc_ggn_feedback_frequency_sweep.py : amplitude and frequency sweeps for testing effect of GGN feedback on a single KC.
• kc_ggn_nofeedback.py : script to simulate KC with no feedback alongside KC with feedback inhibition from GGN.
• pn_kc_ggn_network.py : script to setup and simulate the mushroom body network model (uses config.yaml for parameters).
• pn_output.py : script to setup PN spike trains
• tweak_template.py : script to modify an existing network template (in a data file dumped by an earlier simulation).

*NOTE: Many of these scripts have hard coded absolute paths which
need to be updated for running in a new environment.*

• batch_run_remove_kcs_run.sh : example script for running successive simulations after removing high spiking KCs.
• circular_run : scripts for running running successive simulations after removing high spiking KCs. These scripts read last job ids from a specified file to identify the corresponding data dumps, remove high spiking kcs from those model templates, and run the simulation again until no more highspiking KC is left.
• run_fixed_net_changing_stim.py : example utility script to run simulation of a given network model with changed PN inputs as a subprocess.
• run_fixed_net_with_ig.sh : sample script to run a given network template including IG with different PN inputs.
• run_fixed_network_changing_stim.sh : script to run a fixed network template while changing the PN input pattern.
• run_kc_ggn_feedback_amp_sweep.sh, run_kc_ggn_nofeedback_amp_sweep.sh : scripts to test single KC with and without GGN feedback.
• run_mb_net.sh : script to run mushroom body network model in batch mode.
• run_to_reproduce_ggn_vm_no_seed.sh, run_to_reproduce_ggn_vm.sh, run_to_reproduce_ggn_vm_with_shifting_pn.sh: scripts that run multiple simulations while changing parameters to reproduce realistic GGN voltage trace.
• run_with_ig.sh : script to simulate network model with IG included.

• ggn_voltage_attenuation_vclamp.py : check voltage attenuation along GGN arbor when one branch is voltage clamped.
• run_param_sweep_vclamp.py : test voltage attenuation along GGN arbor with voltage clamp while changing passive properties.

• cellmovie.py : dump a video of rotating neuron in 3D
• displaycell.py : display neuronal morphology from SWC file
• morph3d_*.py : display neuronal morphology in 3D using corresponding module. Use vtk/vispy/matplotlib to display 3D morphology.
• neurograph.py : handle morphology as a graph. This file contains data type definitions and functions to read an SWC file and turn it into a graph (using networkx). There is also GGN specific mapping where I assign custom types to specific branches based on anatomical location.

• nrnutils.py : utilities for handling NEURON model
  • convert a NEURON cell model into a networkx graph
  • insert alpha synapses
  • insert ion channel mechanisms
  • set up recording of Vm
• display_celltemplate.py: 3D visualization of the morphology of a neuron specified in NEURON cell template
• localized_input_output.py : apply synaptic inputs at specified branches. This scripts runs simulation with synchronous synaptic inputs at multiple compartments on specific branches.
• localized_input_output_passive_sweep.py : apply synaptic inputs at specified branches of GGN models with different passive properties.
• nrninit.bat, nrninit.sh : batch file (Windows) and shell script (Linux). Initialize PYTHONPATH to include various model related Python scripts
• staggered_input.py : simulate synaptic input at random compartments in specified branch