mrkrd/thorns

Spike analysis software

thorns and waves

__|________|_____|_|_|___|_____|____||_____|_____|_____|____|_____|___||______
_|_______________________|______|_________|_______|_____|____|__|________|_|__
_____|___|__|_____|_______|____|_________________________|__|_________|_______
___|_______|_____|______|_____|_______|__|___|________|______|___|____________
__|__|_______|_____|__|___|______|________|______|______|_____|_______THORNS__

With thorns you can analyze and display spike trains generated by neurons. It can be useful for the analysis of experimental and simulation data using Python. For example, you can easily calculate peristimulus time histogram (PSTH), interspike time histogram (ISIH), vector strength (VS), entrainment and visualize action potentials with raster plot.

waves is a submodule with some useful signal processing and generation functions, e.g. generate ramped tone, amplitude modulation tone, FFT filter, set level (dB_SPL).

The software was originally developed during my PhD in the group of Werner Hemmert at the TUM. It is oriented towards auditory research, but it could be easily extended.

Usage

Don't forget to check our IPython Notebook DEMO and scripts in the examples directory!

Initialize and load spike trains:

import thorns as th
from thorns.datasets import load_anf_zilany2014

spike_trains = load_anf_zilany2014()

Calculate vector strength:

th.vector_strength(spike_trains, freq=1000)

Raster plot:

th.plot_raster(spike_trains)
th.show()

Generate and plot AM tone:

import thorns.waves as wv

sound = wv.amplitude_modulated_tone(
    fs=48e3,
    fm=100,
    fc=1e3,
    m=0.7,
    duration=0.1,
)

wv.plot_signal(sound, fs=48e3)

wv.show()

You can also browse the API documentation at https://pythonhosted.org/thorns/

Features

• Analyzes and displays spike trains
• Uses pandas.DataFrame as the main data container (spike trains, results)
• Handy signal processing and generating functions: thorns.waves
• Map implementation with various backend (also parallel) and caching: thorns.util.map()
• Dumpdb: quickly dump map()'s results in one script and load from another one: thorns.util.dumpdb(), thorns.util.loaddb()
• Pure Python

Installation

In order to use thorns, you'll need to install the following dependencies first:

• Python (2.7)
• Numpy
• Scipy
• Pandas
• PyTables / tables
• Matplotlib
• py-notify (optional, enables notifications)

Next, type in your command line:

pip install thorns

Contribute

• Open tasks: TODO.org (best viewed in Emacs org-mode)
• Issue Tracker: https://github.com/mrkrd/thorns/issues
• Source Code: https://github.com/mrkrd/thorns

License

The project is licensed under the GNU General Public License v3 or later (GPLv3+).

thorns's Issues

Unclosed files/zombie processes

With python 3.5, using thorns.util.map for multiprocessing generates unclosed files and zombie processes (OSError: [Errno 24] Too many open files).

import thorns as th

def dummy(name):
        print(name)

backend = "multiprocessing"

for i in range(1,100):
        test = th.util.map(func = dummy,
                        space = {"name":"Dummy"},
                        backend = backend,
                        cache = "no",
                        kwargs = {})
        print("Iteration: {}".format(i))

inserting pool.close() into maps._multiprocessing_map() after pooling is done fixes it.

TypeError: 'numpy.float64' object cannot be interpreted as an integer

thorns/thorns/waves.py

Line 171 in 1fdc510

ramp_signal = np.linspace(0, 1, np.ceil(ramp * fs))

Missing cast to int, should be
n_samples = int(np.ceil(ramp * fs))
ramp_signal = np.linspace(0, 1, n_samples )

Python 3.9.7
numpy==1.22.2
thorns==1