Magland, J. et al. Elife 9, e55167 (2020).
Spike sorting involves the extraction of the activity of individual neurons from a sea of activity in extracellular recordings. While manual spike sorting is feasible for small datasets, automated spike sorting algorithms are a necessity for large-scale recordings. Magland et al. have established SpikeForest, a software suite and dedicated website for benchmarking spike sorting algorithms. The researchers ran ten popular spike sorting algorithms on 13 electrophysiology datasets that were associated with ground truth data. While most algorithms excelled for one or more of the datasets, none of the algorithms emerged as a clear winner across all datasets. This result shows that the choice of spike sorting algorithm should be guided by the characteristics of the dataset to be analyzed. For example, MountainSort4 performs well on tetrode and monotrode recordings, while Kilosort2 shines in recordings with low signal-to-noise ratio. Benchmarking results as well as Docker containers for the ten algorithms are available at https://spikeforest.flatironinstitute.org.
Author information
Authors and Affiliations
Corresponding author
Rights and permissions
About this article
Cite this article
Vogt, N. Benchmarked spike sorting. Nat Methods 17, 656 (2020). https://doi.org/10.1038/s41592-020-0902-0
Published:
Issue Date:
DOI: https://doi.org/10.1038/s41592-020-0902-0