(430a) Integrating High-Throughput Data with Biochemical Networks Identifies Functional Regulatory Interactions

Emerging high-throughput technologies allow the rapid measurement of potential regulatory interactions, yet their ability to predict cellular phenotype is still poor, mainly due to the lack of methods to directly connect measured interactions to observable phenotypes, and the presence of a plethora of false-positive interactions. We present an approach – Gene Expression and Metabolism Integrated for Network Inference (GEMINI) – that effectively assesses interactions by using a biochemically-detailed metabolic network to predict the phenotypic consequences of regulation on cellular metabolism. By comparing predictions to phenomics data, GEMINI can produce a refined network model that is significantly better at recalling true interactions than conventional expression-based methods and show increased accuracy in predicting gene knockout phenotypes under different conditions. GEMINI represents an attempt to curate the inference of regulatory interactions by leveraging on a metabolic network—an approach that incorporates multiple layers of biological context and a mechanistic framework to the problem of regulation. 

References

[1]    Chandrasekaran et al, Behavior-Specific Changes in Transcriptional Modules Lead to Distinct & Predictable Neurogenomic States, PNAS, 2011.

[2]    Chandrasekaran & Price, Probabilistic integrative modeling of genome-scale metabolic and regulatory networks in E.coli and M. tuberculosis, PNAS, 2010.

[3]    Barret et al, The global transcriptional regulatory network for metabolism in E.coli exhibits few dominant states, PNAS, 2005.