Netzoor: A Software Infrastructure For The Inference And Analysis Of Gene Regulatory Networks

Author(s): Marouen Ben Guebila, Tian Wang, John Quackenbush

Affiliation(s): Harvard T.H. Chan School of Public Health

Twitter: @marouenbg

The reconstruction of gene regulatory networks requires the development of software tools to integrate data from various genomic modalities. Our research group has developed several network methods to infer biological networks and compare them by conducting differential analyses in a case versus control setting. We grouped these tools in a method ‘zoo’ called netZoo (http://netzoo.gitub.io) where each tool is named after an animal. We harmonized their implementations which allowed us to build interfaces between them and enabled building integrated analytical pipeline. In the latest release, netZooR, the R implementation of netZoo, included PANDA and OTTER to model context-specific gene regulatory networks, EGRET to integrate genetic variants into regulatory network inference, and LIONESS to infer single-sample regulatory networks. NetZooR also includes tools for community detection such as CONDOR and ALPACA to estimate community structures and conduct differential analyses between them. Finally, we developed MONSTER to estimate the transitions between biological state and the regulatory proteins that drive them. NetZoo has been developed as an ecosystem that includes the GRAND database (https://grand.networkmedicine.org) to host gene regulatory network and visualize them on the browser, and Netbooks (http://netbooks.networkmedicine.org) as a Jupyter hub server that includes more than 20 use cases from our previous investigations that address questions in human health and disease. As large-scale genomic characterization of biological systems develops, we will continue to expand netZooR to model genome-scale regulatory networks and build predictive models of gene regulation.