eigenangles/balanced_network

Evaluating network rewiring in balanced random network models

This folder contains the workflow to create and simulate balanced random neural networks, rewire their connectivity by given protocols, compare networks with the eigenangle test, and visualize the results. The network configurations are specified via the config.py file. The following wildcards in the file paths can update the parameters of the config file as in N{N}_f{f}_mu{mu}_p{epsilon}_eta{eta}_sex{sex}_sin{sin}_{syndist} (in the following referred to as {network_specs}). The main functionalities of the workflow are:

• snakemake rewire_and_redraw creates and simulates all (rewired) networks with the parameter ranges specified in the config file.
• snakemake rewire_comparisons compares the original network (weight and correlation matrix) with the rewired networks for each initialization, i.e. random seed.
• snakemake redraw_comparisons compares the different network initializations, i.e. across seeds.
• snakemake results/{network_specs}/rewiring_results.csv processes and aggregates all results in that folder into a dataframe.
• snakemake simulation_output/{network_specs}/seed_{seed}/{protocol}/<weights/correlations/spikes_{t_start}-{t_stop}ms>.png plots the corresponding weight-, correlation matrix, or spike rasterplot.
• snakemake images/eigenspectrum/{network_specs}_{protocol}.png plots the corresponding eigenvalue and eigenangle distributions based on 8 random initializations.
• snakemake images/pvalue_overview/{network_specs}-{protocol}.png plots the p-value swarm plots for weights- and correlations comparisons, their ratio, and the firing rate correlations.
• snakemake images/pvalue_trend/{network_specs}-{protocol}.png plots the p-value trends, their ratio, and the rate correlations for the with respect to the corresponding protocol parameters.