neuropixels-analysis

You can run spike sorting on Neuropixels extracellular recordings by executing the Python pipeline script, specifying your SpikeGLX data path, output directory, sorter, and curation method. This automates preprocessing, sorting, and unit curation end-to-end.

Yes, high-density neural recording analysis supports multiple spike sorting algorithms including Kilosort4, Mountainsort5, SpykingCircus2, and Tridesclous2. You can select your preferred sorter as a command-line argument when running the pipeline.

Unit curation for extracellular electrophysiology data uses threshold-based filtering, model-based refinement with UnitRefine, and AI-assisted visual review. It computes quality metrics like SNR, ISI violations, and isolation distance to assess unit quality.

Yes, SpikeInterface and related libraries like probeinterface and neo are required to preprocess Neuropixels data. The pipeline relies on these dependencies to handle data loading, motion correction, and drift correction.

The best way to analyze high-density neural recordings is using an end-to-end pipeline that handles preprocessing, spike sorting, quality metrics computation, and unit curation. This automates the workflow from raw data loading to publication-ready curated units.

Spike sorting requires drift correction for Neuropixels probes because high-density extracellular recordings often experience electrode drift over time. Applying motion correction ensures accurate spike extraction and reduces false unit merges or splits during sorting.