Guides

In-depth guides covering the main workflows in SpikeLab.

• Loading Data
  • From pickle files
  • From HDF5 files
  • From NWB files
  • From KiloSort/Phy output
  • From SpikeInterface
  • From SpikeLab sorted .npz
  • From the IBL database
  • From Neo SpikeTrains
  • From raw data
• Spike Sorting and Curation
  • Spike Sorting
  • Unit Curation
  • Sorting Concatenated Recordings
  • Handling Sort Failures
• Spike Analysis
  • Population Rate
  • Burst Detection
  • Burst Widths
  • Burst Sensitivity Analysis
  • Per-Unit Metrics
  • Burst Participation per Unit
  • Fraction of Spikes in Bursts
  • Selecting and Combining Data
  • Pairwise Metrics
  • Further Reading
• Instantaneous Firing Rates
  • Instantaneous Firing Rates
  • Dimensionality Reduction
  • Pairwise Firing-Rate Correlations
  • Further Reading
• Event-Aligned Analysis
  • Aligning to Events
  • RateSliceStack
  • SpikeSliceStack
  • Slice-to-Slice Unit Correlations
  • Slice-to-Slice Time Correlations
  • Unit-to-Unit PCA
  • Rank-Order Analysis
  • Temporal Trends and Stability
• Pairwise Analysis
  • Spike Time Tiling Coefficient (STTC)
  • Firing-Rate Correlations
  • FR Correlation vs STTC Scatter
  • Network Analysis
  • Spatial Network Visualization
• Shuffled Data
  • Single Shuffle
  • Shuffle Stacks
  • Unit Subset Stacks
  • Comparing to the Null Distribution
• GPLVM Analysis
  • Fitting a GPLVM
  • Post-fit Analysis
  • Visualisation
• Exporting Data
  • To HDF5
  • To NWB
  • To KiloSort
  • To pickle
• Workspace Persistence
  • Creating a Workspace
  • Saving and Loading
  • Listing and Retrieving
  • Lazy Loading
  • Merging Workspaces
• MCP Server
  • Starting the server
  • Tool categories
  • Workspace addressing
• Parallel Computing
  • Thread-Pool Parallelism
  • Numba Acceleration
• Batch Jobs
  • Prerequisites
  • Setting Up a Session
  • Analysis Jobs
  • Sorting Jobs
  • Policy Guardrails
  • Docker Images
  • Monitoring Jobs
• Cell-Type Classification with HIPPIE
  • Installation
  • Data requirements
  • Pipeline A — Kilosort output + raw .bin file
  • Pipeline B — NWB file with raw traces
  • Pipeline C — Waveforms already available
  • Quick start (waveforms already present)
  • Unsupervised VAE compression (no conditioning)
  • Advanced options
  • Using via the MCP server
  • How it works
  • Checkpoint