neuropixels-analysis

This skill provides a comprehensive toolkit for neuroscientists and researchers working with high-density extracellular electrophysiology data. It facilitates the end-to-end analysis of Neuropixels 1.0 and 2.0 probe recordings, integrating industry-standard libraries such as SpikeInterface, Allen Institute criteria, and International Brain Laboratory (IBL) protocols to ensure reproducible and high-quality results.

The tool is designed for labs conducting large-scale neural recording experiments who need to transition from raw data acquisition to publication-ready curated units. It handles complex preprocessing tasks, such as phase shift correction, bad channel detection, and common reference subtraction, before automating the spike sorting process using high-performance algorithms like Kilosort4, SpykingCircus2, or Mountainsort5.

Key features and capabilities include:

• Seamless data ingestion for SpikeGLX (.ap.bin, .meta), Open Ephys (.continuous), and NWB formats.
• Advanced motion and drift estimation/correction tools tailored for electrophysiology.
• Automated spike sorting pipeline integration with support for multi-processing and GPU acceleration.
• Comprehensive unit quality metric computation including SNR, ISI violations, and presence ratio analysis.
• AI-assisted curation workflows that allow users to leverage LLMs to inspect waveform and correlogram visualizations for expert-level decision making.
• Flexible export options for further analysis in Phy or long-term archiving in the NWB (Neurodata Without Borders) standard.

Usage notes and practical tips:

• Always run motion estimation before spike sorting to ensure data stability and improve sorting accuracy.
• The skill supports interactive visual inspection; when running in environments like Claude Code, users can prompt the agent to analyze waveforms and correlogram plots directly.
• For large datasets, leverage the provided job_kwargs configuration to optimize parallel processing and chunking, preventing memory overflows.
• Use the Allen/IBL presets for standardized, conservative curation to reduce false positives in unit classification.
• The system generates HTML reports and summary logs, providing a transparent audit trail for all preprocessing and curation steps performed during the analysis session.