daniaki/pyPPI

A machine learning tool to assign edges in protein-protein interaction networks with a function.

Introduction

A computational tool for annotating the edges in a protein-protein interaction (PPI) network with post-translational modification (PTM) and functional labels. The tool serves two main purposes:

1. Mining protein-protein interactions labelled with PTMs and higher level signalling function (activation, inhibition etc) from various databases such as KEGG and HPRD.

2. Provide machine learning API that can be used to classify the PTM of binary protein interactions encoded by UniProt accessions to both learn from a training dataset and and classify new/unseen classify interactions.

This can be used as a command line tool with the provided scripts, or custom made scripts. Alternatively, the tool can be integrated into an existing project using the simple API described in the examples. See the documentation for an in-depth guide.

Basic install guide

First clone the repository or download it and extract it to a convenient location. To install the package and all requirements:

python setup.py install

To download all required data:

python setup.py download_data

To test your installation:

python setup.py test 

Or

python test.py

Before using the API or provided scripts, you will need to build the training database and train an initial classifier. Once you have installed the package you can move the script files to any directory of your choice. Once you have changed your working directory to the script directory:

python build_data.py -h
python build_data.py

Documentation

The documentation is not currently hosted (upcoming). To build a local copy of the documentation cd into docs and run the make script:

make html

This will build a html version of the documentation in the folder build. To view the documentation, open the file index.html.