Methods to build a network of pathway co-occurrence relationships out of expression signatures extracted from transcriptomic compendia.
Switch branches/tags
Nothing to show
Clone or download
Fetching latest commit…
Cannot retrieve the latest commit at this time.
Failed to load latest commit information.



Python 3 implementation of methods described in Chen et al.'s 2017 PathCORE-T paper.

Note that this software was renamed from PathCORE to PathCORE-T in Oct 2017. The T specifies that pathway co-occurrence relationships are identified using features extracted from transcriptomic data. The module itself is still named pathcore to maintain backwards compatibility for users of the original PathCORE software package.

This code has been tested on Python 3.5. The documentation for the modules in the package can be accessed here.


To install the current PyPI version (recommended), run:

pip install PathCORE-T

For the latest GitHub version, run:

pip install git+


We recommend that users of the PathCORE-T software begin by reviewing the examples in the PathCORE-T-analysis repository. The analysis repository contains shell scripts and wrapper analysis scripts that demonstrate how to run the methods in this package on features constructed from a broad compendium according to the workflow we describe in our paper.

Specifically, this Jupyter notebook is a simple example of the workflow and a great place to start.

Package contents

The methods in this module are used to identify the pathways overrepresented in features extracted from a transcriptomic dataset of genes-by-samples. Features must preserve the genes in the dataset and assign weights to these genes based on some distribution. [feature_pathway_overrepresentation documentation.]

Contains the data structure CoNetwork that stores information about the pathway co-occurrence network. The output from a pathway enrichment analysis in serves as input into the CoNetwork constructor. [CoNetwork documentation.]

The methods in this module are used to filter the constructed co-occurence network. We implement a permutation test that evaluates and removes edges (pathway-pathway relationships) in the network that cannot be distinguished from a null model of random associations. The null model is created by generating N permutations of the network. [network_permutation_test documentation.]


This work was supported by the Penn Institute for Bioinformatics