Integrative Model of Regulation and Metabolism Improves Growth Phenotype Prediction in Saccharomyces Cerevisiae

The integration of genome-scale gene regulatory networks with metabolic networks into unified predictive models has proven to be a challenging problem – in part because the modeling frameworks for these distinct network types are generally very different.  This challenge has been especially acute for eukaryotes where the gene regulatory processes are typically more complex. The Probabilistic Regulation of Metabolism (PROM) method provided an integration framework, but relied primarily on CHIP-chip/seq or literature information rather than any amount of network inference, and its accuracy in eukaryote networks was significantly less than in bacteria. n this study, we developed a novel method, called Integrated De novo REgulation And Metabolism (IDREAM), by linking a statistically inferred environment and gene regulatory influence network (EGRIN) with a metabolic model using the PROM framework to predict phenotypes and novel genetic interactions associated with mutants of genes encoding transcription factors (TF) in Saccharomyces cerevisiae. The IDREAM model generated predictions of growth that were significantly more correlated with experimentally measured growth than predictions made by the PROM model using regulatory associations from YEASTRACT database. This improvement was consistent whether measured by Pearson or Matthews correlation and under all environmental conditions explored and every metabolic model tested. Furthermore, the IDREAM model predicted, and we experimentally validated, novel genetic interactions involving the genes encoding Oaf1 and members of the pyruvate dehydrogenase (PDH) complex, suggesting a role for Oaf1 in acetyl-CoA regulation in peroxisomes for utilization by the TCA cycle in the absence of a functional PDH complex. The effectiveness of IDREAM for S. cerevisiae can be attributed to the constraints of direct interactions according to the EGRIN false discovery rate and the enhanced ability to distinguish between activator and inhibitor status of TFs. In conclusion, IDREAM provides the first de novo integrated approach to uncover the relationship between gene regulatory interactions and metabolic pathways to predict downstream phenotypes.

---

Extended Abstract: File Not Uploaded

See more of this Session: Biotechnology Applications
See more of this Group/Topical: General Submissions