Integrating epigenetics to construct gene regulatory networks

The biological processes that drive cellular function can be modeled by a complex network of interactions between regulators (transcription factors) and their targets(genes), summarized by gene regulatory networks (GRNs). The cell’s “epigenetic state” governs the potential targeting of genes by influencing chromatin accessibility. However, integrating such information to construct GRNs remains a challenge. Here, we develop an approach SPIDER using epigenetic information (DNase-I Seq data) and message-passing algorithm to estimate networks between transcription factors and genes in multiple cell lines. We validate our predictions against public ChIP-Seq data. SPIDER was more accurate, in predicting GRNs that other methods that integrate epigenetics compared to existing methods and improved detection of cell-line specific interactions in respective GRNs. SPIDER was also able to identify indirect interactions when putative motifs are absent in the regulatory region of genes, but with ChIP-Seq evidence of regulation. The epigenetically-informed GRNs from SPIDER can be used to identify targets of key regulators, or regulators of important genes from an experiment, in the given context of the cell-type.