BeEPy

Behavioral + electrophysiology analyses in python

For a MATLAB equivalent, see CMBHOME (https://github.com/hasselmonians/CMBHOME)

Data structure

The main datastructures are:

• beepy.Unit: Holds spike times, and indexes into Lfp and Session to get variables at spike times

  • Unit.data: A dynamic dictionary of variables (from LFP and behavior) at spike time

• beepy.Lfp: Holds LFP data

  • Lfp.data: A dynamic dictionary of time-varying variables

• beepy.Session: Holds behavioral data, session metadata, a dictionary of Units, and a dictionary of LFPs.

  • Session.data: Dynamic dictionary of behavioral variables
  • Session.active_unit: Key (eg: [0,0]) telling which unit to parse
  • Session.active_lfp: Key (eg: [0]) telling which LFP channel to parse
  • Session.epoch: List of time periods, in seconds, telling what behavioral periods to look at (eg: trials)
  • Session.mode: When running with multiple epochs, should an analysis concatenate into one large epoch (session.mode = 'cat'), or evaluate seperately for each (session.mode = 'map')

• The primary functionality of the datastructures is the introduction of dynamic dictionaries. These fields take some underlying time-varying signal, and index into it based on the active LFP, active Unit, and epoch.

Time variables:

• LFP and Session come with a small number of predefined variables at load time:
  • LFP
    • ts: Time stamp (seconds)
    • signal: Raw trace (mV?)
  • Session
    • x: X position of animal (cm)
    • y: Y position of animal (cm)
    • hd: Head direction (radians)
• Methods exist for adding certain variables:
  • LFP
    • add_band: Adds the filtered signal and instantaneous frequency+phase+magnitude of a specific frequency band
  • Session
    • vel: Calculates the Kalman filtered velocity, angular velocity, acceleration, and angular acceleration
• For any other variable Lfp.add_custom and Session.add_custom allow you to pass a variable with the same length as the underlying timestamps (along with a name for the key). These will be added to the dynamic dictionaries and calculated just as the built-in variables are.

Analyses

• Analyses should accept a Session, respecting any states within it (eg: Session.epoch, Session.active_unit, Session.active_lfp)
• Broken up into submodules based on the nature of the analysis:
  • Analyses.fields: Receptive fields (# spikes / occupancy)
    • Analyses.fields.ratemap: 2D ratemap (default: x & y coordinates)
    • Analyses.fields.headdir: 1D polar ratemap (default: headdir)
  • Analyses.lfp: Phase locking, phase coding, and phase-coupling measures here ...

IO

• Functions for importing from various external datastructures.
• Function for saving to a NeurodataWithoutBorders (.nwb) format.

Installation:

• git clone https://github.com/wchapman/BeEPy.git
• cd BeEPy
• pip install -e .