Figure 1.

Overview of the GRACE algorithm: (A) GRACE integrates data relevant to gene expression and regulation to construct an initial gene regulatory network with transcription factors (red) and target genes (orange) represented as nodes and regulatory links as edges (blue). (B) Co-regulation related datasets are used to construct a meta network with regulatory links now represented as nodes (blue) to extract individual modules (tightly connected regulatory links). (C) An ensemble of Markov Random Fields is trained to re-evaluate and select regulatory links (blue) that form the final high confidence gene regulatory network.