Raw Data Structure and Preprocessing

The raw data are saved at /Raw, with each session organized by recording day and session index (if one day with multiple sessions).
Raw session folders are compressed into a tar.gz file due to limitations of the platform. You can extract them by tools like 7-zip.

Within each session folder, there are:

  • NPX_XXDDDDDD_exp_gx folder – Raw electrophysiology data recorded using SpikeGLX, where XX denotes the subject name and DDDDDD denotes the recording date.
  • .bhv file – Behavioral data recorded with MonkeyLogic (ML) software
  • kilosort_def_5block_97 – Contains the output of Kilosort4
  • LFPprep – Contains the preprocessed LFP signals
  • processed – Processed and converted data formats for downstream analyses.

NPX_ folder contents

The NPX_ directory contains the raw neuronal signal recordings acquired using SpikeGLX, separated into three data streams. Each stream consists of a .bin file and a corresponding .meta file that stores acquisition metadata (e.g., sampling rate, gain, channel map).

Streams:
1. ap – Action potential (spike) data, ~30 kHz sampling rate.
2. lf – Local field potential (LFP) data, 2.5 kHz.
3. ni – Auxiliary task-related inputs (~10 kHz), recorded via a National Instruments card PXI6341, such as event codes sent from the stimulus computer, and analog input from the photodiode.

Inspecting data by eye
You can download SpikeGLX here, and directly open the .bin file with Open File Viewer. Detailed info can be found here. If you don't have a NI card in your PC, you need to use SpikeGLX_NISiM.exe instead of SpikeGLX.exe.

.bhv file

  • Recorded with MonkeyLogic (ML) software, containing stimuli order, eye position, reward time.
  • Can be loaded using the mlread function provided by ML, for convenience, a converted .mat version is also provided in the processed folder
  • When loaded into MATLAB, the data are stored as a structure indexed by ML trial. Note that multiple stimulus onsets (Stimulus Trials) occur within a single ML Trial.
  • Official document of bhv file can be found here.

processed

  • Within each processed folder, there are:
  • BC – A folder containing result of quality metric computed by BombCell.
  • fscale.mat - A scaling factor transforms int16 value into microvolt.
  • ML_YYMMDD_Subject_.mat – Converted MonkeyLogic bhv file into mat.
  • META_YYMMDD_Subject_NSD1000_LOC – Metadata about trial order, image onset time, eye monitor data and metadata from SpikeGLX .meta file. This file is the output of preprocessing script step1.
  • GoodUnitRaw_YYMMDD_Subject_NSD1000_LOC.mat - The raw spike time, quality metric and necessary meta data. This file is the output of preprocessing script step2.
  • GoodUnit_YYMMDD_Subject_NSD1000_LOC_g2 - Goodunit file, the output of preprocessing step3, you can find detailed info here.
  • GoodLFP_YYMMDD_Subject_NSD1000_LOC_g2 - Processed file of local field potential data.
  • Some figure for quality assessment during pre-processing pipeline.