Abstract
Gene and protein interaction experiments provide unique opportunities to study their wiring in a cell. Integrating this information with high-throughput functional genomics data can help identifying networks associated with complex diseases and phenotypes.
Here we propose a unified statistical framework to test network properties of single and multiple genesets. We focused on testing whether a geneset exhibits network properties and if two genesets are strongly interacting with each other.
We then assessed power and false discovery rate of the proposed tests, showing that tests based on a probabilistic model of gene and protein interaction are the most robust.
We implemented our tests in an open-source framework, called Python Geneset Network Analysis (PyGNA), which provides an integrated environment for network studies. While most available tools are designed as web applications, we designed PyGNA to be easily integrated into existing high-performance data analysis pipelines.
Our software is available on GitHub (http://github.com/stracquadaniolab/pygna) and can be easily installed from PyPi or anaconda.
