neuropixels-analysis

End-to-end spike sorting for Neuropixels recordings involves preprocessing raw data, applying motion correction, running GPU-accelerated Kilosort4, and exporting results to Phy. This skill provides scripts to automate this workflow from data loading to curated units for both NP 1.0 and 2.0 probes.

AI-assisted unit curation uses AI-powered visual analysis to help label and evaluate spike sorting output quality. It integrates with quality metrics to streamline decision-making on unit reliability and exports curated labels alongside comprehensive quality reports.

Neuropixels analysis supports both NP 1.0 and NP 2.0 probes for preprocessing, motion correction, and spike sorting. The workflow applies consistently across probe versions to generate analyzers and compute quality metrics.

You need a GPU to run Kilosort4 for spike sorting within this workflow. The pipeline requires GPU acceleration for Kilosort4 alongside SpikeInterface, pandas, matplotlib, numpy, and scipy to process raw data into curated units.

Computing quality metrics after spike sorting with SpikeInterface is handled by postprocessing scripts that generate analyzers and comprehensive reports. The workflow calculates these metrics automatically before integrating AI-assisted curation and exporting results to Phy.

The best way to export spike sorting results for publication review is generating Phy exports and comprehensive quality reports. This workflow produces curated labels and analyzers directly from the processed Neuropixels data to facilitate review.