Alterlab Neuropixels

Neuropixels Data Analysis

Overview

Comprehensive toolkit for analyzing Neuropixels high-density neural recordings using current best practices from SpikeInterface, Allen Institute, and International Brain Laboratory (IBL). Supports the full workflow from raw data to publication-ready curated units.

When to Use This Skill

This skill should be used when:

• Working with Neuropixels recordings (.ap.bin, .lf.bin, .meta files)
• Loading data from SpikeGLX, Open Ephys, or NWB formats
• Preprocessing neural recordings (filtering, CAR, bad channel detection)
• Detecting and correcting motion/drift in recordings
• Running spike sorting (Kilosort4, SpykingCircus2, Mountainsort5)
• Computing quality metrics (SNR, ISI violations, presence ratio)
• Curating units using Allen/IBL criteria
• Creating visualizations of neural data
• Exporting results to Phy or NWB

Does NOT Trigger

Supported Hardware & Formats

Quick Start

Basic Import and Setup

pythonCopy code
import spikeinterface.full as si

# Bundled helper functions live in scripts/neuropixels_pipeline.py
from scripts.neuropixels_pipeline import (
    load_recording, preprocess, check_drift, correct_motion,
    run_spike_sorting, postprocess, curate_units, export_results, run_pipeline,
)

# Configure parallel processing
job_kwargs = dict(n_jobs=-1, chunk_duration='1s', progress_bar=True)

Loading Data

pythonCopy code
# SpikeGLX (most common)
recording = si.read_spikeglx('/path/to/data', stream_id='imec0.ap')

# Open Ephys (common for many labs)
recording = si.read_openephys('/path/to/Record_Node_101/')

# Check available streams
streams, ids = si.get_neo_streams('spikeglx', '/path/to/data')
print(streams)  # ['imec0.ap', 'imec0.lf', 'nidq']

# For testing with subset of data
recording = recording.frame_slice(0, int(60 * recording.get_sampling_frequency()))

Complete Pipeline (One Command)

pythonCopy code
# Run full analysis pipeline (writes all outputs under output_path/)
from scripts.neuropixels_pipeline import run_pipeline

run_pipeline(
    data_path='/path/to/data',
    output_path='output/',
    sorter='kilosort4',
    stream_name='imec0.ap',
    apply_motion_correction=True,
    curation_method='allen',
)

# Results are written to disk:
#   output/sorting_output/    spike sorter output
#   output/analyzer/          SortingAnalyzer (waveforms, metrics)
#   output/quality_metrics.csv
#   output/curation_labels.json

Or run it from the command line:

bashCopy code
python scripts/neuropixels_pipeline.py /path/to/data output/ --sorter kilosort4 --curation allen

Standard Analysis Workflow

1. Preprocessing

pythonCopy code
# Recommended preprocessing chain
rec = si.highpass_filter(recording, freq_min=400)
rec = si.phase_shift(rec)  # Required for Neuropixels 1.0
bad_ids, _ = si.detect_bad_channels(rec)
rec = rec.remove_channels(bad_ids)
rec = si.common_reference(rec, operator='median')

# Or use the bundled wrapper (returns the preprocessed recording + bad channel ids)
from scripts.neuropixels_pipeline import preprocess
rec, bad_channels = preprocess(recording)

2. Check and Correct Drift

pythonCopy code
from scripts.neuropixels_pipeline import check_drift, correct_motion

# Check for drift (always do this!) — detects/localizes peaks and saves
# a drift plot to <output_folder>/drift_check.png, returns a dict with
# 'drift_estimate' (μm range).
drift_info = check_drift(rec, output_folder='output/')

# Apply correction if needed
if drift_info['drift_estimate'] > 20:  # microns
    rec = correct_motion(rec, output_folder='output/', preset='nonrigid_fast_and_accurate')

3. Spike Sorting

pythonCopy code
# Kilosort4 (recommended, requires GPU)
sorting = si.run_sorter('kilosort4', rec, folder='ks4_output')

# CPU alternatives
sorting = si.run_sorter('tridesclous2', rec, folder='tdc2_output')
sorting = si.run_sorter('spykingcircus2', rec, folder='sc2_output')
sorting = si.run_sorter('mountainsort5', rec, folder='ms5_output')

# Check available sorters
print(si.installed_sorters())

4. Postprocessing

pythonCopy code
# Create analyzer and compute all extensions
analyzer = si.create_sorting_analyzer(sorting, rec, sparse=True)

analyzer.compute('random_spikes', max_spikes_per_unit=500)
analyzer.compute('waveforms', ms_before=1.0, ms_after=2.0)
analyzer.compute('templates', operators=['average', 'std'])
analyzer.compute('spike_amplitudes')
analyzer.compute('correlograms', window_ms=50.0, bin_ms=1.0)
analyzer.compute('unit_locations', method='monopolar_triangulation')
analyzer.compute('quality_metrics')

metrics = analyzer.get_extension('quality_metrics').get_data()

5. Curation

pythonCopy code
# Allen Institute criteria (conservative)
good_units = metrics.query("""
    presence_ratio > 0.9 and
    isi_violations_ratio < 0.5 and
    amplitude_cutoff < 0.1
""").index.tolist()

# Or use automated curation (returns {unit_id: 'good'|'mua'|'noise'})
from scripts.neuropixels_pipeline import curate_units
labels = curate_units(metrics, method='allen')  # 'allen', 'ibl', 'strict'

6. AI-Assisted Curation (For Uncertain Units)

When using this skill with Claude Code, Claude can directly analyze waveform plots and provide expert curation decisions. The recommended workflow is to render per-unit summary plots with SpikeInterface and let Claude inspect them:

pythonCopy code
import spikeinterface.widgets as sw
import matplotlib.pyplot as plt

# Find borderline units worth a visual look
uncertain = metrics.query('snr > 3 and snr < 8').index.tolist()

# Render a summary figure per uncertain unit (waveform + correlogram + amplitudes)
for unit_id in uncertain:
    sw.plot_unit_summary(analyzer, unit_id=unit_id)
    plt.savefig(f'ai_curation/unit_{unit_id}_summary.png', dpi=150, bbox_inches='tight')
    plt.close()

Claude Code Integration: When running within Claude Code, ask Claude to examine the saved waveform/correlogram plots directly - no API setup required.

7. Generate Analysis Report

pythonCopy code
# The bundled run_pipeline writes a machine-readable summary.json
# (sampling rate, duration, channel count, unit counts) into output_path/.
import json
with open('output/summary.json') as f:
    summary = json.load(f)
print(summary)

# For a browsable HTML report of waveforms/metrics, use SpikeInterface's exporter:
si.export_report(analyzer, output_folder='output/report/')
# Open output/report/index.html for figures and the per-unit table

8. Export Results

pythonCopy code
# Export to Phy for manual review
si.export_to_phy(analyzer, output_folder='phy_export/',
                 compute_pc_features=True, compute_amplitudes=True)

# Export to NWB (via NeuroConv — SpikeInterface has no native NWB exporter)
# pip install neuroconv
from neuroconv.tools.spikeinterface import write_sorting, write_recording
write_recording(recording=rec, nwbfile_path='output.nwb', overwrite=True)
write_sorting(sorting=sorting, nwbfile_path='output.nwb')

# Save quality metrics
metrics.to_csv('quality_metrics.csv')

Common Pitfalls and Best Practices

1. Always check drift before spike sorting - drift > 10μm significantly impacts quality
2. Use phase_shift for Neuropixels 1.0 probes (not needed for 2.0)
3. Save preprocessed data to avoid recomputing - use rec.save(folder='preprocessed/')
4. Use GPU for Kilosort4 - it's 10-50x faster than CPU alternatives
5. Review uncertain units manually - automated curation is a starting point
6. Combine metrics with AI - use metrics for clear cases, AI for borderline units
7. Document your thresholds - different analyses may need different criteria
8. Export to Phy for critical experiments - human oversight is valuable

Key Parameters to Adjust

Preprocessing

• freq_min: Highpass cutoff (300-400 Hz typical)
• detect_threshold: Bad channel detection sensitivity

Motion Correction

• preset: 'kilosort_like' (fast) or 'nonrigid_accurate' (better for severe drift)

Spike Sorting (Kilosort4)

• batch_size: Samples per batch (30000 default)
• nblocks: Number of drift blocks (increase for long recordings)
• Th_learned: Detection threshold (lower = more spikes)

Quality Metrics

• snr_threshold: Signal-to-noise cutoff (3-5 typical)
• isi_violations_ratio: Refractory violations (0.01-0.5)
• presence_ratio: Recording coverage (0.5-0.95)

Bundled Resources

scripts/preprocess_recording.py

Automated preprocessing script:

bashCopy code
python scripts/preprocess_recording.py /path/to/data --output preprocessed/

scripts/run_sorting.py

Run spike sorting:

bashCopy code
python scripts/run_sorting.py preprocessed/ --sorter kilosort4 --output sorting/

scripts/compute_metrics.py

Compute quality metrics and apply curation:

bashCopy code
python scripts/compute_metrics.py sorting/ preprocessed/ --output metrics/ --curation allen

scripts/export_to_phy.py

Export to Phy for manual curation:

bashCopy code
python scripts/export_to_phy.py metrics/analyzer --output phy_export/

assets/analysis_template.py

Complete analysis template. Copy and customize:

bashCopy code
cp assets/analysis_template.py my_analysis.py
# Edit parameters and run
python my_analysis.py

references/standard_workflow.md

Detailed step-by-step workflow with explanations for each stage.

references/api_reference.md

Quick function reference organized by module.

references/plotting_guide.md

Comprehensive visualization guide for publication-quality figures.

Detailed Reference Guides

Installation

bashCopy code
# Core packages
uv pip install "spikeinterface[full]" probeinterface neo

# External sorters (separate packages, run via si.run_sorter)
uv pip install kilosort            # Kilosort4 (CUDA GPU required)
uv pip install mountainsort5       # Mountainsort5 (CPU)

# Internal sorters ship INSIDE spikeinterface — install the extra, not a
# standalone package. There is no `spykingcircus` package for SpykingCircus2.
uv pip install "spikeinterface[spykingcircus2]"
uv pip install "spikeinterface[tridesclous2]"

# Our toolkit ships as local scripts (scripts/) — no pip install needed;
# run them directly or import from scripts.neuropixels_pipeline

# Optional: AI curation
uv pip install anthropic

# Optional: IBL tools
uv pip install ibl-neuropixel ibllib

SpikeInterface 0.105 — breaking changes worth knowing

0.105.0 (2026-09-21) renamed a lot of surface area. Objects saved by older versions still load, but scripts written against 0.104 may not run:

• peak_sign is gone from the analysis API. The main channel and peak sign are now fixed when the SortingAnalyzer is created and stored as the main_channel_id sorting property, so compute_quality_metrics, compute_template_metrics, compute_spike_amplitudes, compute_spike_locations and compute_unit_locations no longer accept it. compute_snrs(..., peak_sign=, peak_mode=) became compute_snrs(..., method=).
• get_template_extremum_channel() -> SortingAnalyzer.get_main_channels() (old names remain as deprecated shims).
• estimate_sparsity() now defaults to peak_sign="both" instead of "neg" — a silent change in results, not an error, so re-check sparsity-dependent output.
• Probe handling: set_probe()/set_probegroup() act in place and return None; use select_channels_with_probe() to get a new recording. reset_probe() -> remove_probe().
• ChunkRecordingExecutor -> TimeSeriesChunkExecutor; BaseExtractor.load_from_folder() removed in favour of si.load().
• New in 0.105: dartsort as an external sorter, SortingAnalyzer lazy and read_only modes, detect_and_remove_artifacts preprocessing, and read_kilosort4_motion() to build a Motion object from a Kilosort4 output folder.

Pin the SpikeInterface version in any analysis you intend to reproduce, and record it in the methods section — metric definitions have changed across minor releases.

Project Structure

Copy code
project/
├── raw_data/
│   └── recording_g0/
│       └── recording_g0_imec0/
│           ├── recording_g0_t0.imec0.ap.bin
│           └── recording_g0_t0.imec0.ap.meta
├── preprocessed/           # Saved preprocessed recording
├── motion/                 # Motion estimation results
├── sorting_output/         # Spike sorter output
├── analyzer/               # SortingAnalyzer (waveforms, metrics)
├── phy_export/             # For manual curation
├── ai_curation/            # AI analysis reports
└── results/
    ├── quality_metrics.csv
    ├── curation_labels.json
    └── output.nwb

Additional Resources

• SpikeInterface Docs: https://spikeinterface.readthedocs.io/
• Neuropixels Tutorial: https://spikeinterface.readthedocs.io/en/stable/how_to/analyze_neuropixels.html
• Kilosort4 GitHub: https://github.com/MouseLand/Kilosort
• IBL Neuropixel Tools: https://github.com/int-brain-lab/ibl-neuropixel
• Allen Institute ecephys: https://github.com/AllenInstitute/ecephys_spike_sorting
• Bombcell (Automated QC): https://github.com/Julie-Fabre/bombcell
• SpikeAgent (AI Curation): https://github.com/SpikeAgent/SpikeAgent