This code reproduces the modeling results for the paper "Lateral inhibition in V1 controls neural & perceptual contrast sensitivity" by Joseph Del Rosario, Stefano Coletta, Soon Ho Kim, Zach Mobille, Kayla Peelman, Brice Williams, Alan J Otsuki, Alejandra Del Castillo Valerio, Kendell Worden, Lou T. Blanpain, Lyndah Lovell, Hannah Choi, Bilal Haider.
Running simulations require installation of PyNEST (version 3.3.0 used) and NESTML (version 5.1.0). Please note that the current versions of PyNEST and NESTML differ from the ones used in the study, so we recommend using the aforementioned versions. We have provided an Anaconda environment file environment.yml to assist installation:
conda env create -f environment.yml
the above command will create an environment with the correct versions.
NESTML is used to install the dendritic integration model iaf_cond_exp_dend.nestml. To install the model, first (if applicable) follow the directions here to configure settings. In particular, you may need to adjust the ~/anaconda/env/nest/bin/nest-config file with the correct compiler path if you are using a conda environment. Then, use the commands
cd code/custom_model
python install_dendritic.py
to install the model. Some scripts also use matplotlib and scipy for visualization and analysis.
To reproduce Figs. 4C-D, run the distal stimulation protocol for PV and SST stimulation and 5 contrast values. Repeat for 100 seeds (which uses 100 different random seeds for network initialization and input spike generation).
Figs. 4C-D use alpha = 0.07 (strength of dendritic nonlinearity) and pfar_sst = 0.8 (proportion of sst projections which are long-range). To initialize the working directory use the command
python initiallize_parameters.py 0.07 0.8
cd work_dir/base_a0.07_pf0.8
then run
python ../../code/run_distal_stim.py 1 1
where the first argument is the RNG seed and the second is the stimulation condition (vary from 1-10 for each seed). We recommend running simulations in parallel if you intend on running many seeds. After running simulations, the contrast response curves can be plotted using Fig3.ipynb.
Repeat distal stimulations for alpha 0.04-0.07 (0.002 linear spacing) and pf 0-0.8 (0.1 spacing) over 100 seeds. The notebook computes slope changes and plots distributions. Pre-compiled MI changes are stored in compiled_data/Exc_heatmaps.mat.
Plots spatial profile of network activity using simualtions from the above.
Test for inhibition stabilization. Use code/run_delin.py to run simulations with inhibition deleted from the network. Inhibition can be deleted locally or globally by varying the del_range parameter. Initialize using
python initialize_delete_inhibition.py
cd work_dir/dl_0.0
python ../../run_delin.py 1 1
etc.
To produce the data for figure S11, run the run_distal_stim.py script with parameter values pfar_sst = 0.8, alpha = 0.07 in the spontaneous condition and at the contrast values = 0.05 for both PV and SST stimulation (30 network seeds). To produce the figures, run the notebook FigS11.ipynb
To produce the data for figure S12, run the scripts run_local_stim.py, run_0p25_stim.py, and run_distal_stim.py over the parameter values pfar_sst = 0.2, 0.5, 0.8 and alpha = 0.04, 0.05, 0.06, 0.07, 0.08 with 10 different network seeds across contrast values = 0.02, 0.05, 0.1, 0.18, 0.33. To produce the plots, run the jupyter notebook FigS12.ipynb
To produce the data for figure S13, run the run_distal_stim.py script with parameter values pfar_sst = 0.2, 0.8 and alpha = 0.07 across contrast values = 0.02, 0.05, 0.1, 0.18, 0.33 for both PV and SST stimulation protocols (9 network seeds). To produce the plots, run the notebook FigS13.ipynb
To produce the data for figure S22, run the script run_varyDist_stim.py with pfar_sst = 0.8, alpha = 0.08 over 11 network seeds across contrast values = 0.02, 0.05, 0.1, 0.18, 0.33. The fourth argument to run_varyDist_stim.py is the laser position, which should range from 0 to 0.5. We use 11 evenly-spaced values in Fig. S22. To produce the plots, run the notebook FigS22.ipynb
To produce the data for figure S24, run run_distal_stim_noAnti.py with pfar = 0.2, 0.8, alpha = 0.07 over 9 network seeds across contrast values = 0.02, 0.05, 0.1, 0.18, 0.33 for both PV and SST stimulation protocols. To produce the plots, run the notebook FigS24.ipynb