blechpy 2.0.65

Package for exrtacting, processing and analyzing Intan and OpenEphys data

See the full documentation here.

• blechpy
• Installation
• Usage
• Datasets
  • Starting wit a raw dataset
    • Create dataset
    • Initialize Parameters
    • Basic Processing
    • Viewing a Dataset
  • Loading an existing dataset
  • Import processed dataset into dataset framework
• Experiments
  • Creating an experiment
  • Editing recordings
  • Held unit detection

Table of contents generated with markdown-toc

blechpy

This is a package to extract, process and analyze electrophysiology data recorded with Intan or OpenEphys recording systems. This package is customized to store experiment and analysis metadata for the BLECh Lab (Katz lab) @ Brandeis University, but can readily be used and customized for other labs.

Installation

I recommend installing miniconda to handle your virtual environments Create a miniconda environment with:

conda env create --name blechpy python==3.7
conda activate blechpy

Now you can install this package simply with pip:

pip install blechpy

Now you can deal with all of your data from within an ipython terminal: ipython

import blechpy

Usage

blechpy handles experimental metadata using data_objects which are tied to a directory encompassing some level of data. Existing types of data_objects include:

• dataset
  • object for a single recording session
• experiment
  • object encompasing an ordered set of recordings from a single animal
  • individual recordings must first be processes as datasets
• project
  • object that can encompass multiple experiments & data groups and allow analysis or group differences

Datasets

Right now this pipeline is only compatible with recordings done with Intan's 'one file per channel' or 'one file per signal type' recordings settings.

Starting wit a raw dataset

Create dataset

With a brand new shiny recording you can initilize a dataset with:

dat = blechpy.dataset('path/to/recording/directory')
# or
dat = blechpy.dataset()  # for user interface to select directory

This will create a new dataset object and setup basic file paths. If you're working via SSH or just want a command-line interface instead of a GUI you can use the keyword argument shell=True

Initialize Parameters

dat.initParams() 
# or
dat.initParams(shell=True)

Initalizes all analysis parameters with a series of prompts. See prompts for optional keyword params. Primarily setups parameters for:

• Flattening Port & Channel in Electrode designations
• Common average referencing
• Labelling areas of electrodes
• Labelling digital inputs & outputs
• Labelling dead electrodes
• Clustering parameters
• Spike array creation
• PSTH creation
• Palatability/Identity Responsiveness calculations

Initial parameters are pulled from default json files in the dio subpackage. Parameters for a dataset are written to json files in a parameters folder in the recording directory

Basic Processing

dat.processing_status

Can provide an overview of basic data extraction and processing steps that need to be taken.

An example data extraction workflow would be:

dat = blechpy.dataset('/path/to/data/dir/')
dat.initParams()            # See fucntion docstring, lots of optional parameters to eliminate need for user interaction
dat.extract_data()          # Extracts raw data into HDF5 store
dat.create_trial_list()     # Creates table of digital input triggers
dat.mark_dead_channels()    # View traces and label electrodes as dead, or just pass list of dead channels
dat.common_average_reference() # Use common average referencing on data. 
                               # Repalces raw with referenced data in HDF5 store
dat.detect_spikes()
dat.blech_clust_run()       # Cluster data using GMM
dat.blech_clust_run(data_quality='noisy') # re-run clustering with less strict parameters

dat.sort_units(electrode_number)        # Split, merge and label clusters as units

Viewing a Dataset

Experiments can be easily viewed wih: print(dat) A summary can also be exported to a text with: dat.export_to_text()

Loading an existing dataset

dat = blechpy.load_dataset() # load an existing dataset from .p file
# or
dat = blechpy.load_dataset('path/to/recording/directory') 
# or
dat = blechpy.load_dataset('path/to/dataset/save/file.p')

Import processed dataset into dataset framework

dat = blechpy.port_in_dataset()
# or
dat = blechpy.port_in_dataset('/path/to/recording/directory')

Experiments

Creating an experiment

exp = blechpy.experiment('/path/to/dir/encasing/recordings')
# or
exp = blechpy.experiment()

This will initalize an experiment with all recording folders within the chosen directory.

Editing recordings

exp.add_recording('/path/to/new/recording/dir/')    # Add recording
exp.remove_recording('rec_label')                   # remove a recording dir 

Recordings are assigned labels when added to the experiment that can be used to easily reference exerpiments.

Held unit detection

exp.detect_held_units()

Uses raw waveforms from sorted units to determine if units can be confidently classified as "held". Results are stored in exp.held_units as a pandas DataFrame. This also creates plots and exports data to a created directory: /path/to/experiment/experiment-name_analysis

Analysis

The blechpy.analysis module has a lot of useful tools for analyzing your data. Most notable is the blechpy.analysis.poissonHMM module which will allow fitting of the HMM models to your data. See tutorials.