The documentation is here: http://tridesclous.readthedocs.io/
Authors: Christophe Pouzat and Samuel Garcia
tris des clous is a very dishonest translation of spike sorting to French.
Pronouce it [tree day clue] in English.
The primary goal of tridesclous is to provide a toolkit to teach good practices in spike sorting techniques. This tools is now mature and can be used for experimental data.
The forest of spike sorting tools is dense and tridesclous is a new tree. Be curious and try it.
tridesclous is a complete rewrite of our old (and unsuccessful) tools (SpikeOmatic, OpenElectrophy, ...) with up-to-date (in 2017) python modules to simplify our codes.
- should make it easy to keep a trace of your spike sorting process in a jupyter notebook.
- offer a UI written in Qt for interactive exploration.
- can be used for online spikesorting with pyacq. See online_demo.py
- tridesclous is quite fast. For a tetrode dataset, you can expect X30 speedup over real time on a simple laptop.
- some pieces of algorithm are written both in pure python (numpy/scipy/...) and OpenCL (filter, peak detetion, ...). So tridesclous is efficient for large array (>=64 channel).
- each piece of the algorithm is written with chunk by chunk in mind. So the offline tridesclous is not agressive for memory.
- tridesclous used neo for reading dataset. So many format are available (Raw, Blackrock, Neuralynx, Plexon, Spike2, Tdt, ...)
- tridesclous is open source and based on true opensource stack.
- some datasets are available for testing it now here https://github.com/tridesclous/tridesclous_datasets