neuropy Goto Github PK

neuropy is a Python application for interactive analysis of neuronal spike, LFP, and stimulus data.

The user-facing data object hierarchy looks like this:

                                level:

                Animal            1
                  |
                Track             2             Track
                  |                               |
              Recording           3           TrackSort
             /    |    \                          |
          Sort   LFP  Experiment  4          TrackNeuron
           |
         Neuron                   5

Once data are loaded, they can be accessed in various ways:

animal.track.recording
animal.tr[trackid].r[recordingid]
recording.n[neuronid]

For example, to plot a population raster plot of all active cells in recording 5 of track 1 of animal example1, you can type:

example1.tr1.r05.praster()

Different object types have different analysis methods associated with them. Thanks to IPython, tab completion should help you discover analysis methods, and typing ? after the method name or simply typing the opening parenthesis should give you a docstring.

As for internal variable scope, this is how things work:

Main Qt event loop creates a
    NeuropyWindow which runs an
        IPython session which imports
            neuropy modules

So, something like get_ipython() only works from within those neuropy modules that have been imported during the IPython session, but not in any of NeuropyWindow's methods.

Some more details are available in Appendix C.3 of my thesis.

Dependencies:

neuropy requires recent versions of the following to be installed:

• Python (3.x, 2.x no longer supported)
• IPython 5.3.0
• numpy
• scipy
• matplotlib
• PyQt4 (PySide will probably work too, but is untested)
• Cython

Optional:

• scikit-learn

To ensure that the Qt v2 API is used by IPython, run export QT_API=pyqt at the command line before running neuropy, or add it to your .bashrc or .bash_aliases file.

Some analyses assume the use of custom matplotlib default settings in the included matplotlibrc file.

neuropy is developed in Xubuntu 14.04. It should work in other Linux distributions. In principle, it should also work in Windows and OSX.

An older version is described in the paper [Python for large-scale electrophysiology] (http://www.frontiersin.org/Neuroinformatics/10.3389/neuro.11.009.2008/abstract).

To run:

$ python main.py # in the neuropy folder

To install for use as a library:

$ python setup.py install