NetREX: Network Rewiring using EXpression - Towards Context Specifc Regulatory Networks

Authors

• Y Wang
• D Cho
• H Lee
• J Fear
• B Oliver
• T Przytycka

Abstract

Gene regulatory networks (GRNs) describe regulatory relationship between transcription factors and their target genes. Methods to infer GRNs are typically context-agnostic. However, many practical applications require context-specific networks. We introduce NetREX (Network Rewiring using Expression), a novel and powerful method that given a prior network that is related to the target network and context-specific expression data, constructs a context-specific GRN by rewiring the prior network. Complementing NetREX, we developed PriorBoost, a method to gauge the similarity of the prior network to the (unknown) target network and thus to detect instances when the prior network might not be sufficiently related. Unlike other prior-based methods which often end up constructing a network that is further away from the true network than the initial prior network, NetREX is the first method that provides consistent improvement over the prior. This property was achieved due to several important biological and methodological insights. We used NetREX to infer sex-specific GRNs for Drosophila using a previously reported condition agnostic GRN as the prior. Using PriorBoost, we demonstrated that the previously reported network was female biased. Consequently, it provided an excellent prior for reconstructing the female network but could only be helpful for reconstructing the male network after excluding the testis specific genes. We evaluated NetREX using the gold standard network in E. coli. For the sex-specific Drosophila GRNs, we used two novel scoring functions which were validated to be reliable in a situation when a gold standard network is not available, and Doublesex (DSX) target genes for validation.

Scientific Focus Area: Computational Biology

This page was last updated on Friday, March 26, 2021