Skip to content

Files

Latest commit

32e1960 · May 7, 2018

History

History
72 lines (36 loc) · 4.18 KB

README_old.md

File metadata and controls

72 lines (36 loc) · 4.18 KB

MountainLab Spike Sorting Software

About

MountainSort (a component of MountainLab) is spike sorting software developed by Jeremy Magland, Alex Barnett, and Leslie Greengard at the Center for Computational Biology, Flatiron Institute in close collaboration with Jason Chung and Loren Frank at UCSF department of Physiology. It is part of MountainLab, a general framework for data analysis and visualization.

MountainLab software is being developed by Jeremy Magland and Witold Wysota.

The software comprises tools for processing electrophysiological recordings and for visualizing and validating the results.

Contact the authors for information on the slack team for users and developers.

Announcement: As of November 21st, 2017, a beta release of the new version of MountainSort/MountainLab is available here: https://github.com/flatironinstitute/mountainsort. Please use that version going forward.

Installation

Please use the new version of MountainSort and MountainLab (see the announcement below). But here are the old installation instructions.

How to run spike sorting

Please use the new version of MountainSort and MountainLab (see the announcement below). But here are the old instructions for the first sort

Working branches

Some demo videos

Tests

Repo of unit (and not so unit) tests

Data formats used in MountainLab

The .mda file format

Data management

The .prv data management system

Automated curation/annotation

Documentation on using annotation scripts will be forthcoming.

Because one of the goals of mountainsort is to enable reproducible spike sorting, we strongly advise against manual modifications that go beyond merging bursting clusters and perhaps rejecting certain noise clusters. Instead, we suggest that you export the cluster metrics along with the sorted clusters and then set cutoffs for inclusion of data in analyses based on those metrics. This will make it easy to describe your subsequent analyses and easy to determine how those analyses do or do not depend on cluster quality.

The isolation and noise overlap metrics we provide work well for the situations we have focused on, but they can be supplemented or replaced by other objective metrics as needed. Such metric processors may be included in the pipeline as post-processing plugins as C++, matlab, or python modules. Contact us if you you would like to contribute additional cluster metrics, or need help with integration.

Other documentation

A guide to using MountainSort with snippets, rather than continuous data acquisition

An old guide: Cluster metrics and automated curation

Acknowledgements

Thanks to all the users on the slack team for ongoing testing and feedback.

References

Barnett, Alex H., Jeremy F. Magland, and Leslie F. Greengard. "Validation of Neural Spike Sorting Algorithms without Ground-truth Information." Journal of Neuroscience Methods 264 (2016): 65-77. Link to arXiv

Magland, Jeremy F., and Alex H. Barnett. Unimodal clustering using isotonic regression: ISO-SPLIT. Link to arXiv