Analysis and prediction of morphological and electrophysiological features of neurons using model trained using experimental data.
These instructions will get you a copy of the project up and running on your local machine for development and testing purposes. As of now this is only a prototype and only the dev version is available.
To run this software you need several libraries:
- numpy >= 1.13
- scipy >= 0.19.1
- networkx >= 1.11
- scikit-learn >= 0.19.0
- matplotlib >= 2.0.2
- pandas >= 0.20.3
- seaborn >= 0.8
- pygraphviz >= 1.3.1
These libraries come with the Anaconda bundle (www.anaconda.com) and for Linux users can be obtained by calling:
<sudo> apt-get install anaconda
In case of Windows users an .exe installer is avalailable at (www.anaconda.com/download/). For a manual install of each library the user can execute:
conda install <library>
or for cases when libraries are not available through the conda channel:
pip install <library>
Once these dependencies are installed the library can be cloned from the GitHub repository:
git clone https://github.com/mllera14/multi-output-regression <destination-folder>
To install it open a console in and type:
python setup.py install
The library can now be imported into any development enviornment as:
import electromorpho as emorph
A notebook with examples on how to use the library is included in the repository (see example_notebook.ipynb)