Home
Lussac is an automated and configurable analysis pipeline for post-processing and merging multiple spike-sorting analyses.
The goal is to improve the yield and quality of data from multielectrode extracellular recordings by merging the outputs of different spike-sorting algorithms and/or multiple analysis runs with different parameters.
For more information, check out our preprint.
When we started development of Lussac, our goal was to improve processing of cerebellar complex spikes, which are a challenging problem with extracellular recordings of the cerebellar cortex.
This resulted in a principled method for automating post-processing and for combining multiple analyses from different packages and/or using different parameters. The methods developed proved beneficial to spike sorting generally, not just for cerebellar recordings.
Lussac is still in a Beta version. We are working on an improved version (using the newest versions of SpikeInterface, ProbeInterface and more intuitive use, improved modules...).
For now, most tests have been done on Ubuntu 20.04 LTS on cerebellar datasets (and synthetic datasets) with 64 channels or less. Lussac theoretically works with datasets containing more channels (e.g. Neuropixels), but some optimisations might be required (you can contact us or create an issue if you have some problems).
We would be delighted if you test the software and welcome feedback/requests for help. We aim to make the software as easy to use as possible.
We tested our algorithm using a synthetic data set simulating cortical pyramidal cell and interneurone activity (Jun et al. 2017) available through SpikeForest. Since all spike times of all neurones are known, we could easily compare our algorithm to standard runs of a single spike-sorting algorithms in terms of recovery and contaminating events.
