# Hladnik & Grewe: Receptive field sizes and neuronal encoding bandwidth are constrained by axonal conduction delays.

This repository contains raw data of neuronal activity in P-type electroreceptors of the electric fish *Apteronotus leptorhynchus*, derived data i.e. analysis results, the python scripts running the analyses, simulations and the plotting.

## ``raw_data`` folder

This folder contains the raw datasets. Data and metadata are stored in the open [NIX](https://github.com/g-node/nix) container format. To read these data files you need the [``nixio``](https://github.com/g-node/nixpy) python library. For a more comfortable approach use the [``rlxnix``](https://github.com/relacs/relacsed_nix) python package.
Usually, the files contain the raw membrane potential recording, the detected action potentials, the animal's electric organ discharge (EOD) together with the respective metadata. *Note:* Some datasets are older and have been used in previously analyses. The dataset structure deviates a bit from newer files.

## ``derived_data`` folder

This folder contains the analysis results. The contained files will be overwritten when re-running the analysis scripts. *csv-files are written with ``pandas`` using the semicolon as delimiter. To read them use something like the following pattern ``df = pd.read_csv(filename, sep=";", index_col=0)``.
*npz-files are compressed ``numpy`` files.

## ``stimuli`` folder

Contains a single file, which is the white noise stimulus waveform.

## ``file_lists`` folder

Two simple text files that are used by the analysis scripts: 
1. listing the datasets in which the receptive field was measured, and from which we extract the baseline response properties.
2. The datasets in which white-noise responses were measured.

## ``figures`` folder

Contains some png-files which will be embedded into the final figures. The folder is the default output folder for created figures.

## ``code`` package

python files performing the analyses, simulations and plotting. Three sub-packages:

1. ``analyses``: The data analysis scripts working on the raw data
2. ``plots``: Plotting scripts
3. ``simulations``: LIF model simulations

## Scripts

To run the analyses, simulations, or plot the figures, you can use one of the three scripts

1. ``run_analyses.py`` command line tool for running all data analyses. (**Note: analysis of the heterogeneous populations will take a while!**)
2. ``run_simulations.py`` command line tool to run the LIF simulations. (**Note: This will take a while!**)
3. ``plot_figures.py`` command line script that allows to plot the figures of the publication.

The first two take only a single optional argument which is the number of parallel processes that should be spawned for the respective task. E.g.
```
python3 run_simulations.py --help  # to show the help text
python3 run_simulations.py -j 12  # run the simulations using 12 parallel processes
```

The plot tool expects as a first argument the sub command (i.e. figure) that will be run. Each of these sub-commands define a few more arguments. Using the default arguments should work.

```
python3 plot_figures.py --help  # to list all available subcommands
```

For example:
```
python3 plot_figures.py figure1 -h  # to show the options for figure 1
python3 plot_figures.py figure1 -n  # to plot figure 1 and show it.
```

If the ``-n`` argument is given, the figure will be shown but not saved. If the option is omitted, the figure will be saved to the ``figures`` folder, but not shown.
The default arguments should lead to the same figures as in the paper.

## Dependencies

The scripts depend on the following packages (installed version):
* python (3.8.10)
* numpy (1.19.2)
* pandas (1.1.0)
* scipy (1.7.0)
* nixio (1.5.2)
* rlxnix (0.6.10, for [installation options see](https://github.com/relacs/relacsed_nix))
* joblib (0.16.0)
* sklearn (0.23.1)
* matplotlib (3.4.2)