P-sort

P-sort performs semi-automated spike sorting of electrophysiological recordings from cerebellar Purkinje cells (P-cells) to distinguish and attribute complex spikes and simple spikes for cerebellar neurophysiology analyses.

Key Features:

• Semi‑Automated Spike Sorting: Detects and attributes spikes in P-cell recordings, distinguishing variable complex spikes from simple spikes and reducing misidentification and misattribution.
• Nonlinear Dimensionality Reduction: Applies advanced nonlinear dimensionality reduction techniques to cluster spike waveforms and accommodate variability in complex spike shapes.
• Statistical Relationship Utilization: Leverages statistical relationships between simple and complex spikes to merge disparate clusters or split single clusters to improve sorting accuracy.
• Performance Benchmarking: Demonstrated improved complex-spike identification and reduced misattribution relative to three automatic sorters in comparative studies of over 300 P-cells from marmosets, macaques, and mice, occasionally surpassing expert manual curation.
• Labeled Dataset: Includes provision of labeled data from 313 recording sessions containing statistical characteristics of waveforms and firing patterns.

Scientific Applications:

• Cerebellar Neurophysiology: Enables accurate identification and attribution of complex and simple spikes to support studies of Purkinje cell functional dynamics.
• Cross‑Species Electrophysiology: Supports comparative analyses of P-cell firing patterns across marmosets, macaques, and mice through validated sorting and benchmarking.

Methodology:

Nonlinear dimensionality reduction combined with statistical analysis of complex and simple spike waveforms.