Neuropixels recording datasets associated with the manuscript "Conjoint specification of action by neocortex and striatum"

Data description

Each mat file (‘js2p0_tbyt…’) contains structures: ‘jkvt’, ‘ss’, ‘spikeTimesCell’, and ‘trI’. 

• The structure ‘ss’ would be most relevant at least for our initial work. Thus, I’ll explain each field in ‘ss’. 
• ‘ss’ consists of 31 fields, of which length equals to the total number of trials in the recording session (e.g., 193 total trials in ‘WR40_081919’).
• • The field ‘trialType’ classifies trials into: 
  • • ‘sp’: a success trial. 
    • ‘ps’: an error trial (error made by pushing instead of pulling the joystick). 
    • ‘pmpp’: an error trial (error made by a premature response before the cue). 
    • ‘to’: a time out trial (no reach was made within the set time limit). 
  • ‘timeAlign’ indicates time in milliseconds to which the neural and behavioral data are aligned to. 
  • ‘evtAlign’ indicates the type of event to which data are aligned to. 
  • • A success trial is aligned to the reach start (‘rStart’).  
    • Other error trials (‘ps’ and ‘pmpp’) are also aligned to ‘rStart’. 
    • A time out trial is aligned to the cue onset (‘trJsReady’).  
  • ‘rEnd’ indicates the time each trial ended by completing a pull (or an error, time out). 
  • ‘spkTimeBins’ indicates time bins used to bin the neural and behavioral data. 
  • • The width of the time bin is 20 ms. 
    • The range is -1 to 2 sec (20 ms x 150 bins) relative to ‘timeAlign’.  
  • ‘unitTimeBCtx’ is the unit-by-bin spike count matrix of the motor cortex.  
  • ‘unitTimeBStr’ is the spike count matrix of the dorsal striatum. 
  • ‘hTrjBfull’ is the full binned hand trajectory on X (left-right), Y (forward-backward) and Z (upward-downward) axes.
  • ‘hTrjB’ is the binned hand trajectory (in millimeters) of which bins are aligned to the ‘spkTimeBins’. Note that they are cut off at the pull stop (or commitment of an error), thus there are fewer bins than that of the spike data. However, the existing bins are temporally aligned. 
  • ‘hVelB’ is the binned hand velocity (in cm/s).  
  • ‘hInitPos’ is just the estimated initial hand position (median amongst the initial bins).   
  • ‘hDistFromInitPos’ is the estimated distance of each bin from the ‘hInitPos’. 
  • ‘maxPullForce’ is the maximum force generated in the pulling direction (negative) of each trial. Empty for error or timeout trials.   
  • ‘forceB’ is the amount of force generated at each time bin. 
  • ‘tPullStart’ is the time when the pull started. 
  • ‘tPullStop’ is the time when the pull stopped. 
  • ‘spkPullIdx’ is the logical index marking spike time bins corresponding to pulling. 
  • ‘spkRchIdx’ is the logical index marking spike time bins corresponding to reaching. 
  • ‘rchSpeed1ms’ is the reach speed (cm/s) estimated at 1-ms time resolution.
  • ‘maxRchSpeed’ is the maximum speed (cm/s) of hand movement of  the trial.
  • ‘maxRchSpeedXYZ’ is the maximum speed (cm/s) of hand movement per each axis. 
  • ‘hTrjBfullP1’ is the full hand trajectory (not aligned with the spike matrix) upto the point of the pull start. 
  • ‘jTrjB’ is the binned joystick trajectory aligned with the spike matrix (‘unitTimeB*’) and hand trajectories (‘hTrjB’).  
  • ‘jsXYbot’ is the joystick position at the target location. 
  • ‘rchAngDeg’ is the reach angle estimated on the 2-d (X-Y) trajectories from the reach onset to the pull onset. 
  • ‘tLaserStart’ is the time of laser onset (if any), that silences the activity of many M1 neurons.  
  • ‘tLaserStop’ is the time of laser offset. 
  • ‘spkTimeBlaserI’ is the logical index indicating spike time bins with the laser on.  
  • ‘blNumber’ indicates the block ID. There are up to 8 blocks. 
  • • Block#1&5: Left position/Low torque
    • Block#2&6: Left position/High torque
    • Block#3&7: Right position/Low torque
    • Block#4&8: Right position/High torque
  • ‘blType’ describes the type of each block. 
  • ‘blShiftLogic’ is just the logical index indicating trials where a block shift occurred. 
• The structure ‘jkvt’ contains more raw behavioral variables, I’ll provide details as needed in the future. 
• The structure ‘spikeTimesCell’ contains information about each recorded unit and their spike times. This will be useful if we want to generate PSTHs with different parameters.  

Code

• This is the git repository with all the code we used for data preprocessing and analysis. We will provide any details as needed.  https://github.com/jup36/MatlabNeuralDataPipeline